# LumaBrowser documentation LumaBrowser is a desktop app that bundles a web browser with a full local AI stack: an LLM runtime that fits models to the user's GPUs, chat with tools and agents, local image, video, music and voice generation, browser automation for external AI agents (MCP, WebDriver, CDP), and triggers that turn browser events into webhooks and automations. Everything runs on the user's own machine. Every code file has a doc here at the same path. When you change a file, update its doc, and update this index if its purpose changed. ## core/shared Small building blocks shared by more than one subsystem. - [core/shared/text/TokenEstimator.md](core/shared/text/TokenEstimator.md): estimates token counts at one shared characters-per-token ratio; has `estimateTokens`, `tokensToChars`. - [core/DebugLog.md](core/DebugLog.md): dev-only ring buffer of console output for "Copy Logs"; has `install`, `isInstalled`, `dump`, `clear`. - [core/shared/AgentNotices.md](core/shared/AgentNotices.md): per-conversation queue of background notices for the agent loop; has `enqueue`, `drain`, `drainAll`, `pending`, `beginRun`, `drainIfSoleRun`, `endRun`, `reset`. - [core/shared/AgentNoticeRun.md](core/shared/AgentNoticeRun.md): handle for one registered agent run; has `drain`, `end`. - [core/shared/AppPaths.md](core/shared/AppPaths.md): picks the base dir for managed models and runtimes and migrates it out of the install dir; has `resolveBaseDir`, `migration`. - [core/shared/ImageModelPicker.md](core/shared/ImageModelPicker.md): quality-first image model pick against one card's budget; has `pickImageModel`, `cardBudgetBytesFromHw`. - [core/shared/ModelDescriptor.md](core/shared/ModelDescriptor.md): one host-independent shape for the LLM and image model lists; has `toLlmDescriptor`, `toImageInstalledDescriptor`, `toImageCatalogDescriptor`. - [core/shared/hardwareTiers.json.md](core/shared/hardwareTiers.json.md): hardware tier rows shared with the website; data file. - [core/shared/audio/WavCodec.md](core/shared/audio/WavCodec.md): dependency-free WAV reader and writer; has `parse`, `encode`, `durationSec`. - [core/shared/audio/PcmSamples.md](core/shared/audio/PcmSamples.md): Float32/Int16 PCM conversion and linear resampling; has `floatToInt16Bytes`, `int16BytesToFloat`, `resampleLinear`. - [core/shared/content/HtmlToMarkdown.md](core/shared/content/HtmlToMarkdown.md): HTML to model-legible Markdown, main content only; has `convert`. - [core/shared/content/HtmlTokenizer.md](core/shared/content/HtmlTokenizer.md): HTML string to open/close/text tokens; has `tokenize`. - [core/shared/content/HtmlTreeBuilder.md](core/shared/content/HtmlTreeBuilder.md): tokens to a forgiving node tree with implied end tags; has `build`. - [core/shared/content/HtmlNode.md](core/shared/content/HtmlNode.md): element or text node of the converter tree; has `attribute`, `find`, `findAll`, `plainText`, `remove`. - [core/shared/content/HtmlElementCatalog.md](core/shared/content/HtmlElementCatalog.md): element categories (void, raw-text, non-content, chrome, block); has `isNonContent`, `isBlock`, `headingLevel`. - [core/shared/content/NodeTextStats.md](core/shared/content/NodeTextStats.md): per-element text and link-text totals; has `measure`, `of`, `linkDensity`. - [core/shared/content/TreePruner.md](core/shared/content/TreePruner.md): base for top-down subtree removal passes; has `prune`, `shouldDrop`. - [core/shared/content/InvisibleContentPruner.md](core/shared/content/InvisibleContentPruner.md): drops metadata, media, controls and hidden elements; has `shouldDrop`. - [core/shared/content/ContentRootLocator.md](core/shared/content/ContentRootLocator.md): picks main, a lone article, or the body; has `locate`. - [core/shared/content/BoilerplatePruner.md](core/shared/content/BoilerplatePruner.md): drops chrome landmarks, roles, widgets and link-dense blocks; has `prune`. - [core/shared/content/UrlResolver.md](core/shared/content/UrlResolver.md): href to absolute http(s) URL against the page base; has `forDocument`, `resolve`. - [core/shared/content/MarkdownWriter.md](core/shared/content/MarkdownWriter.md): Markdown output buffer that owns spacing; has `text`, `lineBreak`, `blockBreak`, `block`. - [core/shared/content/MarkdownRenderer.md](core/shared/content/MarkdownRenderer.md): walks the tree and writes Markdown; has `render`, `renderBlocks`, `renderInline`, `renderChildren`. - [core/shared/content/BlockRenderer.md](core/shared/content/BlockRenderer.md): base for structured block renderers; has `matches`, `render`. - [core/shared/content/CodeBlockRenderer.md](core/shared/content/CodeBlockRenderer.md): pre to fenced code with a language hint; has `render`, `fenceFor`. - [core/shared/content/ListBlockRenderer.md](core/shared/content/ListBlockRenderer.md): ul/ol/menu to Markdown lists; has `render`. - [core/shared/content/QuoteBlockRenderer.md](core/shared/content/QuoteBlockRenderer.md): blockquote to quote lines; has `render`. - [core/shared/content/TableBlockRenderer.md](core/shared/content/TableBlockRenderer.md): data tables to GFM, layout tables unwrapped; has `render`. - [core/shared/content/HtmlEntityDecoder.md](core/shared/content/HtmlEntityDecoder.md): decodes HTML character references; has `decode`. - [core/shared/content/MarkdownTableRenderer.md](core/shared/content/MarkdownTableRenderer.md): renders cell rows as a GFM table; has `render`. - [core/shared/content/ImageDimensions.md](core/shared/content/ImageDimensions.md): PNG/JPEG header dimensions and aspect fit to an area budget; has `read`, `fitToBudget`. - [core/shared/content/SelectorQuerySanitizer.md](core/shared/content/SelectorQuerySanitizer.md): strips session query strings from learned selectors; has `sanitize`. - [core/shared/download/ResumableDownload.md](core/shared/download/ResumableDownload.md): starts or resumes a single-file download, parallel or sequential, with Hub auth, expected size and sha256, free-space guard and backoff; has `download`. - [core/shared/download/ParallelDownload.md](core/shared/download/ParallelDownload.md): concurrent Range-chunk download resumable from a sidecar; has `execute`. - [core/shared/download/SequentialDownload.md](core/shared/download/SequentialDownload.md): single-stream fallback with Range-append resume; has `execute`. - [core/shared/download/ChunkMeta.md](core/shared/download/ChunkMeta.md): crash-safe sidecar recording finished chunks; has `pathFor`, `plan`, `load`, `save`, `clear`, `exists`. - [core/shared/download/DownloadHeaders.md](core/shared/download/DownloadHeaders.md): total size and trusted sha256 from response headers; has `totalFromHeaders`, `sha256FromHeaders`. - [core/shared/download/DownloadRateMeter.md](core/shared/download/DownloadRateMeter.md): smoothed transfer rate and ETA; has `sample`. - [core/shared/download/DownloadVerifier.md](core/shared/download/DownloadVerifier.md): size and sha256 check before rename, typed integrity errors, discards file and chunk map; has `verifyPartial`, `sha256File`. - [core/shared/download/PartialFile.md](core/shared/download/PartialFile.md): `.partial` naming and sizing; has `pathFor`, `size`. - [core/shared/download/HubAuth.md](core/shared/download/HubAuth.md): the HF token source and Hub-host-only Authorization header; has `token`, `isHubUrl`, `headersFor`. - [core/shared/download/DownloadError.md](core/shared/download/DownloadError.md): typed download failures (gated repo, not found, integrity, no space, unsafe path); has `fromStatus`, `fromRequest`, `retryAfterMs`. - [core/shared/download/DownloadRetry.md](core/shared/download/DownloadRetry.md): capped backoff with Retry-After, never on auth, 404 or integrity; has `run`, `shouldRetry`, `delayFor`. - [core/shared/download/SafeRepoPath.md](core/shared/download/SafeRepoPath.md): the one place a remote file name becomes a local path; has `check`, `fileName`, `join`, `isSafe`. - [core/shared/download/FreeSpaceGuard.md](core/shared/download/FreeSpaceGuard.md): refuses a download the volume cannot hold, before any byte lands; has `assert`, `format`. - [core/shared/download/VolumeFreeSpace.md](core/shared/download/VolumeFreeSpace.md): free and total bytes of the volume under a folder, read through the nearest existing ancestor; has `read`, `nearestExisting`. - [core/shared/download/DownloadPublish.md](core/shared/download/DownloadPublish.md): fsync the finished file, rename, fsync the directory; has `publish`. - [core/shared/fs/ContainedPath.md](core/shared/fs/ContainedPath.md): proves a path is strictly inside a root; has `isWithin`, `resolveWithin`, `isImmediateChild`. - [core/shared/fs/FileLinker.md](core/shared/fs/FileLinker.md): places a model file by hardlink, then symlink, never copy; has `link`, `isSameFile`. - [core/shared/ipc/IpcEnvelope.md](core/shared/ipc/IpcEnvelope.md): wraps IPC handlers in the `{ success }` envelope; has `enveloped`, `raw`, `errorFields`, `isCloneable`. - [core/shared/ipc/PathPicker.md](core/shared/ipc/PathPicker.md): native open dialog with one normalised cancel result; has `pick`. - [core/shared/ipc/SenderStream.md](core/shared/ipc/SenderStream.md): progress emitter that stops once the renderer is gone; has `create`, `observe` (main-process observers of every streamed message). - [core/shared/ipc/VoiceChannels.md](core/shared/ipc/VoiceChannels.md): voice IPC channel names; has `VOICE_EVENT_CHANNEL`, `TTS_STREAM_CHANNEL`. - [core/shared/llm/BalancedJson.md](core/shared/llm/BalancedJson.md): finds where a JSON object really ends; has `extractObject`. - [core/shared/llm/KvCacheModes.md](core/shared/llm/KvCacheModes.md): KV cache precision table; has `mode`, `pair`, `idFor`, `shortLabel`. - [core/shared/llm/ModelOverrides.md](core/shared/llm/ModelOverrides.md): the settings a model or a model profile may override and how sparse override layers stack; has `clean`, `merge`, `isEmpty`. - [core/shared/llm/OpenAiChatBody.md](core/shared/llm/OpenAiChatBody.md): single builder for OpenAI-compatible chat bodies; has `build`. - [core/shared/llm/ReasoningEffort.md](core/shared/llm/ReasoningEffort.md): the thinking dial; has `normalizeDial`, `normalizeEffort`, `extraFor`, `resolveDial`, `nearestEffort`, `dialPositionsFor`. - [core/shared/lumabyte/LumaRequestSigner.md](core/shared/lumabyte/LumaRequestSigner.md): lumabyte.com rolling token and identity headers; has `computeToken`, `buildAuthHeaders`. - [core/shared/lumabyte/SignedLumaClient.md](core/shared/lumabyte/SignedLumaClient.md): signed lumabyte.com HTTP client with an egress policy; has `request`, `headerValue`. - [core/shared/net/IpClass.md](core/shared/net/IpClass.md): IP range taxonomy with inbound trust and outbound safety checks; has `classifyIp`, `isLanPeer`, `isConnectableTarget`, `isLoopbackIp`, `normalizeIp`, `clientIp`. - [core/shared/runtime/BinaryLookup.md](core/shared/runtime/BinaryLookup.md): finds a runtime binary inside an extracted archive; has `findBinaryIn`, `pathExists`. - [core/shared/runtime/RuntimeCatalog.md](core/shared/runtime/RuntimeCatalog.md): platform-keyed accessors over a runtime declaration (base class); has `getCatalog`, `getById`, `getAssetPattern`, `getBinaryNames`, `fingerprint`, `hashDeclaration`. - [core/shared/runtime/CudaSnapshot.md](core/shared/runtime/CudaSnapshot.md): decides whether a failed CUDA probe was transient; has `isTransientCudaFailure`. - [core/shared/runtime/FreePort.md](core/shared/runtime/FreePort.md): free loopback ports and the port-window table; has `findFreePort`, `isPortFree`, `ephemeralPort`, `portWindow`. - [core/shared/runtime/HotswapRamGate.md](core/shared/runtime/HotswapRamGate.md): decides whether RAM can hold a hotswap pool; has `evaluate`, `ramReserveBytes`. - [core/shared/runtime/NvidiaSmi.md](core/shared/runtime/NvidiaSmi.md): resolves and runs nvidia-smi asynchronously; has `setSmiPath`, `onPathChange`, `query`, `streamArgs`, `queryGpus`, `queryComputeApps`. - [core/shared/runtime/NvidiaSmiOutputParser.md](core/shared/runtime/NvidiaSmiOutputParser.md): pure nvidia-smi output parsers, cards keyed by index and bus id; has `parseGpuCount`, `parseGpuRows`, `parseGpuRow`, `parseComputeApps`. - [core/shared/runtime/NvidiaSmiStream.md](core/shared/runtime/NvidiaSmiStream.md): the one long-lived nvidia-smi query loop with cached snapshot, backoff restarts and one-shot fallback; has `start`, `stop`, `snapshot`, `refresh`, `whenReady`, `computeApps`, `cachedComputeApps`, `getState`. - [core/shared/runtime/nvidia-smi/NvidiaSmiOutcome.md](core/shared/runtime/nvidia-smi/NvidiaSmiOutcome.md): classifies an nvidia-smi run as ok, missing, timeout, exit or parse; has `fromError`, `fromExit`, `isOk`, `isDegraded`. - [core/shared/runtime/nvidia-smi/NvidiaSmiBlockReader.md](core/shared/runtime/nvidia-smi/NvidiaSmiBlockReader.md): chunked loop stdout into blocks of rows keyed by card index; has `feed`, `flush`, `end`. - [core/shared/runtime/nvidia-smi/RestartBackoff.md](core/shared/runtime/nvidia-smi/RestartBackoff.md): exponential restart delay with cap and reset; has `next`, `peek`, `reset`. - [core/shared/runtime/nvidia-smi/ComputeAppsGate.md](core/shared/runtime/nvidia-smi/ComputeAppsGate.md): re-queries the per-process VRAM list only when a card's used bytes moved; has `changed`, `record`, `signature`. - [core/shared/runtime/ProcessMemory.md](core/shared/runtime/ProcessMemory.md): reads a process's RSS; has `rssBytes`, `rssBytesSync`. - [core/shared/runtime/RuntimeUpdateChecker.md](core/shared/runtime/RuntimeUpdateChecker.md): checks GitHub for newer runtime releases; has `check`. - [core/shared/runtime/SysdepsChecker.md](core/shared/runtime/SysdepsChecker.md): Linux system-library check via ldconfig and ldd; has `checkKnownLibs`, `checkBinary`, `preflight`. - [core/shared/runtime/SysdepsReport.md](core/shared/runtime/SysdepsReport.md): builds and merges library-check results; has `aptLineFor`, `describeMissing`, `build`, `merge`. - [core/shared/runtime/SysdepsLinkerParser.md](core/shared/runtime/SysdepsLinkerParser.md): parses ldconfig and ldd output; has `parseLdconfig`, `parseLddNotFound`. - [core/shared/runtime/SysdepsCommandRunner.md](core/shared/runtime/SysdepsCommandRunner.md): runs a probe command that always resolves; has `run`. - [core/shared/runtime/server/IdleTimer.md](core/shared/runtime/server/IdleTimer.md): idle-unload timer for runtime supervisors; has `set`, `arm`, `disarm`. - [core/shared/runtime/rampin/RamPinWorker.md](core/shared/runtime/rampin/RamPinWorker.md): entry script for the RAM-pin utility process; has `main`. - [core/shared/runtime/rampin/RamPinSession.md](core/shared/runtime/rampin/RamPinSession.md): one RAM-pin worker lifetime; has `handle`. - [core/shared/runtime/rampin/RamPinMessagePort.md](core/shared/runtime/rampin/RamPinMessagePort.md): the RAM-pin worker's link to its parent; has `fromProcess`, `on`, `post`. - [core/shared/runtime/rampin/MemoryLock.md](core/shared/runtime/rampin/MemoryLock.md): base class for platform memory locks; has `prepare`, `mapFile`, `lockAsync`. - [core/shared/runtime/rampin/WindowsMemoryLock.md](core/shared/runtime/rampin/WindowsMemoryLock.md): VirtualLock through koffi; has `prepare`, `mapFile`, `lockAsync`. - [core/shared/runtime/rampin/LinuxMemoryLock.md](core/shared/runtime/rampin/LinuxMemoryLock.md): mlock through koffi with a memlock limit check; has `prepare`, `mapFile`, `lockAsync`. - [core/shared/text/LineDiff.md](core/shared/text/LineDiff.md): bounded line diff into hunks with context; has `diff`. - [core/shared/text/Slug.md](core/shared/text/Slug.md): filesystem- and id-safe slugs; has `from`. - [core/shared/registry/ContributionRegistry.md](core/shared/registry/ContributionRegistry.md): base class for registries extensions contribute to, tracking which extension owns which ids; has `register`, `unregister`, `unregisterByExtension`, `get`, `getById`, `has`, `list`, `extensions`. - [core/shared/HardwareTiers.md](core/shared/HardwareTiers.md): classifies a machine into a hardware tier row; has `classify`, `minGpuBytes`. - [core/shared/llm/ContextBudget.md](core/shared/llm/ContextBudget.md): context-window shares, fixed costs and a coherence check; has `resolveBudget`, `assertCoherent`, `headroomShare`, `compactionTrigger`. - [core/shared/runtime/CudaPin.md](core/shared/runtime/CudaPin.md): picks the CUDA devices a managed server is pinned to; has `resolveCudaDevice`, `filterDiagnosticsToDevices`, `applyCudaDeviceEnv`. - [core/shared/runtime/CudaDeviceProbe.md](core/shared/runtime/CudaDeviceProbe.md): reads CUDA cards from diagnostics or the nvidia-smi stream's cached snapshot, never spawning; has `readDevices`, `withLiveMemory`, `readComputeApps`, `gpuCount`, `whenReady`. - [core/shared/runtime/CudaDevicePicker.md](core/shared/runtime/CudaDevicePicker.md): the most-powerful-free-card pick order; has `pick`, `cardRoomBytes`. - [core/shared/runtime/GpuHeadroom.md](core/shared/runtime/GpuHeadroom.md): the user's per-card VRAM headroom every placement and the VRAM watchdog share; has `bytes`, `load`, `save`. - [core/shared/runtime/ManagedRuntimeSettings.md](core/shared/runtime/ManagedRuntimeSettings.md): saved runtime detection and manual binary paths; has `getManualRuntimeBinary`, `setManualRuntimeBinary`, `getCachedRuntimesView`, `ensureRuntimesView`. - [core/shared/runtime/rampin/RamPinService.md](core/shared/runtime/rampin/RamPinService.md): supervises one RAM-pin worker for a feature's default model; has `apply`, `stop`, `getStatus`, `isActiveFor`. - [core/shared/runtime/rampin/RamPinFitGate.md](core/shared/runtime/rampin/RamPinFitGate.md): refuses a pin that would cut into the OS RAM reserve; has `evaluate`. - [core/shared/runtime/rampin/RamPinWorkerLauncher.md](core/shared/runtime/rampin/RamPinWorkerLauncher.md): forks the RAM-pin worker by path; has `fork`. - [core/shared/runtime/RuntimeInstaller.md](core/shared/runtime/RuntimeInstaller.md): base class for managed runtime installers; has `installRuntime`, `uninstallRuntime`, `fetchLatestRelease`, `resolvePrerelease`. - [core/shared/runtime/RuntimeDetector.md](core/shared/runtime/RuntimeDetector.md): base class for binary runtime detectors; has `detectRuntimes`, `parseVersionOutput`, `readVersion`. - [core/shared/runtime/RuntimeManifest.md](core/shared/runtime/RuntimeManifest.md): the managed runtime's manifest.json; has `build`, `write`, `read`. - [core/shared/runtime/install/GithubRuntimeInstall.md](core/shared/runtime/install/GithubRuntimeInstall.md): one GitHub-release runtime install end to end; has `execute`. - [core/shared/runtime/install/AcquisitionGuard.md](core/shared/runtime/install/AcquisitionGuard.md): refuses early when there is no auto-download path; has `entryFor`, `assertKind`, `assetPatternOrThrow`. - [core/shared/runtime/install/RuntimeInstallError.md](core/shared/runtime/install/RuntimeInstallError.md): install error carrying a code and detail. - [core/shared/runtime/install/GithubReleaseClient.md](core/shared/runtime/install/GithubReleaseClient.md): GitHub release feed and nightly pointers; has `latestRelease`, `latestPrerelease`, `newestReleaseWithAsset`. - [core/shared/runtime/install/ReleaseAssetResolver.md](core/shared/runtime/install/ReleaseAssetResolver.md): picks release and asset by channel; has `resolve`. - [core/shared/runtime/install/AssetDownload.md](core/shared/runtime/install/AssetDownload.md): streams a release asset and returns its sha256; has `toFile`. - [core/shared/runtime/install/ArchiveExtractor.md](core/shared/runtime/install/ArchiveExtractor.md): extracts archives with tar, extract-zip or PowerShell; has `extract`. - [core/shared/runtime/install/SharedLibLinker.md](core/shared/runtime/install/SharedLibLinker.md): links .so files beside the binary on Linux; has `consolidate`. - [core/shared/runtime/install/InstallEvents.md](core/shared/runtime/install/InstallEvents.md): safe install progress emitter; has `emitter`, `releaseFields`, `assetFields`. - [core/shared/runtime/install/ManagedDir.md](core/shared/runtime/install/ManagedDir.md): managed and staging runtime paths; has `pathFor`, `stagingPathFor`, `ensure`, `remove`. - [core/shared/runtime/install/CompanionAssets.md](core/shared/runtime/install/CompanionAssets.md): companion archives (cudart, NCCL) installed on top; has `install`. - [core/shared/runtime/detect/VersionProbe.md](core/shared/runtime/detect/VersionProbe.md): --version probe that never caches a loader error; has `read`, `normalize`. - [core/shared/runtime/detect/LlamaBuildNumber.md](core/shared/runtime/detect/LlamaBuildNumber.md): reads the llama.cpp build number; has `parse`. - [core/shared/runtime/detect/PathBinaryLookup.md](core/shared/runtime/detect/PathBinaryLookup.md): PATH lookup with PATHEXT; has `find`. - [core/shared/runtime/detect/RuntimeHardwareCheck.md](core/shared/runtime/detect/RuntimeHardwareCheck.md): hardware readiness for CUDA, Vulkan, AMD or Intel GPUs (with Intel's compute runtime for SYCL and OpenVINO) and Apple Silicon; has `evaluate`. - [core/shared/runtime/detect/BackendDeviceCheck.md](core/shared/runtime/detect/BackendDeviceCheck.md): asks an installed GPU build which devices its backend loaded (`--list-devices`), with the Linux ldd explanation; has `run`, `parse`. - [core/shared/runtime/detect/IntelComputeRuntime.md](core/shared/runtime/detect/IntelComputeRuntime.md): which of Intel's Linux compute pieces are installed (OpenCL ICD, Level Zero, NPU driver); has `probe`, `PACKAGES`. - [core/shared/runtime/detect/BinaryRuntimeProbe.md](core/shared/runtime/detect/BinaryRuntimeProbe.md): finds a runtime binary (registered, managed dir, PATH); has `detect`. - [core/shared/runtime/detect/ExtensionRuntimeProbe.md](core/shared/runtime/detect/ExtensionRuntimeProbe.md): fail-soft extension runtime detect hook; has `detect`. - [core/shared/runtime/detect/RuntimeDetailFields.md](core/shared/runtime/detect/RuntimeDetailFields.md): stamps entry and host fields (host labels, spawn env) onto a detection row and folds in the backend device check; has `stamp`, `applyBackend`. - [core/shared/runtime/BaseRuntimeServer.md](core/shared/runtime/BaseRuntimeServer.md): base supervisor for one native inference child; has `start`, `stop`, `getStatus`, `setIdleTimeout`, `markActive`, `holdIdle`, `waitUntilSettled`. - [core/shared/runtime/server/ChildReaper.md](core/shared/runtime/server/ChildReaper.md): kills a child and resolves on its real exit; has `reap`, `killNow`. - [core/shared/runtime/server/RuntimeLogRing.md](core/shared/runtime/server/RuntimeLogRing.md): bounded child output log with the stderr tail; has `capture`, `snapshot`, `clear`, `errorTail`. - [core/shared/runtime/server/RuntimeSpawnEnv.md](core/shared/runtime/server/RuntimeSpawnEnv.md): spawn cwd and env (library path, CUDA pin, runtime extra env) for a runtime binary; has `build`, `isPinned`. - [core/shared/runtime/HotswapCoordinator.md](core/shared/runtime/HotswapCoordinator.md): single-occupancy eviction per hotswap pool; has `shared`, `configure`, `register`, `acquire`, `release`, `snapshot`. - [core/shared/runtime/placement/VramReleaseWaiter.md](core/shared/runtime/placement/VramReleaseWaiter.md): waits for a card's VRAM to be reclaimed after eviction; has `read`, `wait`. - [core/shared/runtime/PlacementLayout.md](core/shared/runtime/PlacementLayout.md): unified placement layout schema, normalizer and queries; has `emptyLayout`, `normalizeLayout`, `resourceById`, `orderedDevices`, `placedItems`, `hotswapPools`. - [core/shared/runtime/placement/PlacementStore.md](core/shared/runtime/placement/PlacementStore.md): loads and saves the placement layout; has `load`, `save`, `migrateLegacy`. - [core/shared/runtime/placement/PlacementResolver.md](core/shared/runtime/placement/PlacementResolver.md): resolves an item to ordered devices and RAM; has `resolve`, `deviceRoomMap`. - [core/shared/runtime/placement/RemoteDeviceRef.md](core/shared/runtime/placement/RemoteDeviceRef.md): peer GPU references; has `isRef`, `parse`, `format`. - [core/shared/runtime/VramCoordinator.md](core/shared/runtime/VramCoordinator.md): first-come-first-serve VRAM ledger and multi-GPU placement; has `shared`, `reserve`, `hold`, `release`, `setSplit`, `snapshot`. - [core/shared/runtime/vram/VramLedger.md](core/shared/runtime/vram/VramLedger.md): VRAM claims and per-card debit math; has `set`, `get`, `delete`, `debitMap`, `applyDebit`. - [core/shared/runtime/vram/AutoVramPlacement.md](core/shared/runtime/vram/AutoVramPlacement.md): automatic GPU placement rules; has `place`, `cudaDeviceFor`. - [core/shared/runtime/vram/VramOverride.md](core/shared/runtime/vram/VramOverride.md): per-role explicit CUDA device setting; has `read`, `devices`. - [core/shared/runtime/vram/StateChangeHook.md](core/shared/runtime/vram/StateChangeHook.md): safe supervisor state-change subscription; has `on`. - [core/shared/net/LoopbackRequestGuard.md](core/shared/net/LoopbackRequestGuard.md): refuses DNS-rebinding and cross-site browser requests to local automation servers (Host must be loopback or the bind host; Origin absent or same-port loopback); has `refusal`, `isAllowed`, `rejectUpgrade`, `hostnameOf`. - [core/shared/ipc/IpcSubscription.md](core/shared/ipc/IpcSubscription.md): subscribe functions for preload IPC channels; has `of`, `payload`. ## core/browser - [core/browser/Accelerators.md](core/browser/Accelerators.md): maps a tab keyDown input to a browser action id; has `match`. - [core/browser/ActionEvidence.md](core/browser/ActionEvidence.md): turns before/after page fingerprints into action evidence; has `compare`, `stateKey`, `isFingerprint`. - [core/browser/ActionEvidenceScripts.md](core/browser/ActionEvidenceScripts.md): in-page fingerprint and settle-observer script sources; has `beforeExpr`, `beforeScript`, `pollScript`, `installScript`, `finalScript`. - [core/browser/PageSettler.md](core/browser/PageSettler.md): waits for a page to go quiet and runs page scripts with a timeout; has `settle`, `runBounded`. - [core/browser/ActionWatcher.md](core/browser/ActionWatcher.md): watches one input action (navigations, new tabs) and builds its evidence; has `finish`, `cancel`, `state`. - [core/browser/BrowserActions.md](core/browser/BrowserActions.md): the one browser action table and the REST, MCP and chat surfaces built from it; has `ACTIONS`, `restRoutes`, `mcpToolDefinitions`, `chatToolEntries`, `chatActionIds`. - [core/browser/BrowserService.md](core/browser/BrowserService.md): the browser API extensions get as `context.browser`; has tab and page actions, `attachCDP`, `registerWebviewPreload`, tab event hooks, `emit`. - [core/browser/ChromeOverlay.md](core/browser/ChromeOverlay.md): display-only view layers above the tab views so popups paint over the page; has `show`, `measure`, `setActive`, `hide`, `raiseVisible`, `destroy`. - [core/browser/ChromeObjectShim.md](core/browser/ChromeObjectShim.md): page source that fills `window.chrome` like stock Chrome; has `SOURCE`. - [core/browser/DataFileRender.md](core/browser/DataFileRender.md): shows CSV/TSV downloads as an inline HTML page; has `isRenderableDataFile`, `buildDataFileHtml`. - [core/browser/DownloadManager.md](core/browser/DownloadManager.md): follows Chromium downloads for the downloads shelf; has `track`, `list`, `get`, `cancel`, `open`, `showInFolder`, `clearFinished`. - [core/browser/FaviconCache.md](core/browser/FaviconCache.md): LRU cache of one favicon per hostname saved to JSON; has `hostOf`, `get`, `getMany`, `getForUrl`, `record`, `flush`. - [core/browser/FaviconFetcher.md](core/browser/FaviconFetcher.md): turns a favicon source into a size-capped data URL; has `toDataUrl`. - [core/browser/InputDriver.md](core/browser/InputDriver.md): trusted mouse clicks via sendInputEvent; has `trustedClick`, `toDip`. - [core/browser/LlmFallbackOrchestrator.md](core/browser/LlmFallbackOrchestrator.md): finds a working selector from a plain-language description; has `resolve`, `execute`. - [core/browser/PasskeyShim.md](core/browser/PasskeyShim.md): page source that holds passkey requests the user did not start; has `SOURCE`. - [core/browser/PermissionManager.md](core/browser/PermissionManager.md): deny-by-default web permission policy with camera/microphone prompts; has `install`, `attachSession`, `handleRequest`, `check`, `decideMedia`, `respond`, site list methods. - [core/browser/PermissionKinds.md](core/browser/PermissionKinds.md): permission request origin, media kinds and prompt wording; has `originOf`, `kindsOf`, `kindsOfCheck`, `describe`, `hostOf`. - [core/browser/SitePermissionStore.md](core/browser/SitePermissionStore.md): remembered allow/block answers per origin; has `answerFor`, `remember`, `list`, `clear`, `clearAll`. - [core/browser/SearchEngines.md](core/browser/SearchEngines.md): omnibox search engines and their query URLs; has `byId`, `urlFor`, `currentId`. - [core/browser/TabPreviewManager.md](core/browser/TabPreviewManager.md): shows a live tab inside another tab's renderer rect; has `attach`, `setRect`, `detach`, `focus`, `captureFrame`, `reapply`, `destroy`. - [core/browser/TabPreviewGeometry.md](core/browser/TabPreviewGeometry.md): rect mapping and fit-zoom maths for the tab preview; has `toWindowRect`, `clampZoom`, `widthZoom`, `fittedZoom`. - [core/browser/TabUrl.md](core/browser/TabUrl.md): the single definition of what a tab can load; has `normalize`, `isNavigable`, `isLoopbackOrPrivateHost`. - [core/browser/extraction/ElementDigest.md](core/browser/extraction/ElementDigest.md): page script that lists visible interactive elements and tags them; has `SCRIPT`. - [core/browser/extraction/RowExtract.md](core/browser/extraction/RowExtract.md): page source for reading one repeating element into a row; has `TEXT_OF_SRC`, `EXTRACT_ROW_SRC`. - [core/browser/extraction/SelectorKit.md](core/browser/extraction/SelectorKit.md): spots machine-generated class names and ids; has `isHashedToken`, `HASHED_TOKEN_SRC`. - [core/browser/models/ApiResponse.md](core/browser/models/ApiResponse.md): the JSON envelope every browser REST route replies with; has `success`, `error`, `toJSON`. - [core/browser/vision/CoordinateSpace.md](core/browser/vision/CoordinateSpace.md): converts points between CSS px, screenshot pixels and model grids; has `frameScale`, `imageToCss`, `cssToImage`, `modelToSent`, `sentToModel`, `smartResize`, `zoomWindow`. - [core/browser/vision/GroundingReplyParser.md](core/browser/vision/GroundingReplyParser.md): reads a point or box out of a grounding model reply; has `parse`, `stripReasoning`. - [core/browser/vision/GroundingProfiles.md](core/browser/vision/GroundingProfiles.md): per-family grounding prompt, coordinate format and image size; has `getProfile`, `profileForModel`. - [core/browser/vision/ImageOps.md](core/browser/vision/ImageOps.md): nativeImage size, resize, crop and encode in true pixels; has `normalize`, `size`, `resize`, `crop`, `toDataUrl`, `pngSize`. - [core/browser/widgets/OptionMatcher.md](core/browser/widgets/OptionMatcher.md): decides which select or listbox option the caller meant; has `pickOption`, `describeOptions`, `normalize`. - [core/browser/widgets/DateFormatter.md](core/browser/widgets/DateFormatter.md): turns an ISO date into a native or masked input value; has `parseIsoDate`, `nativeDateValue`, `detectDateFormat`, `formatDate`. - [core/browser/widgets/SliderValue.md](core/browser/widgets/SliderValue.md): clamps and step-snaps a slider value; has `snap`. - [core/browser/BrowserRoutes.md](core/browser/BrowserRoutes.md): /api/browser REST controller derived from BrowserActions; has `create`. - [core/browser/ChromeIdentity.md](core/browser/ChromeIdentity.md): single source of the Chrome identity spoof; has `USER_AGENT`, `CHROME_VERSION`, `UA_METADATA`, `acceptLanguage`, `applyToSession`, `overrideParams`. - [core/browser/identity/ChromeBrandList.md](core/browser/identity/ChromeBrandList.md): Chromium's GREASE brand list; has `build`, `toHeader`. - [core/browser/identity/WindowsPlatformVersion.md](core/browser/identity/WindowsPlatformVersion.md): real sec-ch-ua-platform-version from the OS build; has `detect`, `fromBuild`. - [core/browser/identity/ClientHintHeaders.md](core/browser/identity/ClientHintHeaders.md): rewrites Sec-CH-UA headers to match the spoof; has `rewrite`. - [core/browser/identity/IdentityOverride.md](core/browser/identity/IdentityOverride.md): CDP identity override per webContents and child targets; has `apply`. - [core/browser/LlmFallbackService.md](core/browser/LlmFallbackService.md): natural-language description to CSS selector; has `tryDeterministicResolve`, `validateSelector`, `snapshotFor`, `resolveSelector`. - [core/browser/llm-fallback/SelectorPrompt.md](core/browser/llm-fallback/SelectorPrompt.md): the selector-resolver prompt; has `build`, `snapshotContext`. - [core/browser/llm-fallback/SelectorReplyParser.md](core/browser/llm-fallback/SelectorReplyParser.md): model reply text to one selector; has `parse`. - [core/browser/llm-fallback/SelectorValidation.md](core/browser/llm-fallback/SelectorValidation.md): live-DOM selector uniqueness probe; has `script`, `verdict`. - [core/browser/llm-fallback/EmptyReplyDiagnostic.md](core/browser/llm-fallback/EmptyReplyDiagnostic.md): debug summary of an empty completion; has `summarize`, `fullDump`. - [core/browser/NetworkInterceptor.md](core/browser/NetworkInterceptor.md): per-tab CDP network watching with opt-in body capture; has `attachToWebContents`, `syncFetchInterception`, `getRequestLog`, `detachAll`. - [core/browser/network/RequestLog.md](core/browser/network/RequestLog.md): per-tab URL-only request ring buffer that fails closed; has `add`, `recordResponse`, `forTab`, `forget`. - [core/browser/network/WatcherForwarder.md](core/browser/network/WatcherForwarder.md): redacted per-watcher webhook payloads; has `forward`, `buildPayload`. - [core/browser/ResolutionCachePage.md](core/browser/ResolutionCachePage.md): in-page capture and validate scripts for the resolution cache; has `captureAtPointScript`, `captureForSelectorScript`, `validateScript`. - [core/browser/extraction/SemanticTreeScript.md](core/browser/extraction/SemanticTreeScript.md): flat actionable-element extraction script; has `SOURCE`. - [core/browser/extraction/StructuralSummaryScript.md](core/browser/extraction/StructuralSummaryScript.md): compacted DOM skeleton extraction script; has `SOURCE`. - [core/browser/vision/GroundingClient.md](core/browser/vision/GroundingClient.md): screenshot plus description to a pixel point; has `pass`, `locate`, `sentSizeFor`. - [core/browser/vision/VisionPageScripts.md](core/browser/vision/VisionPageScripts.md): set-of-marks overlay and hit-test scripts; has `drawMarksScript`, `removeMarksScript`, `hitTestScript`. - [core/browser/widgets/KeyInput.md](core/browser/widgets/KeyInput.md): trusted key, type, clear and drag input; has `trustedKey`, `trustedType`, `trustedClearField`, `trustedDrag`. - [core/browser/widgets/WidgetScript.md](core/browser/widgets/WidgetScript.md): shared frame and helpers for widget page scripts; has `wrap`, `refArg`. - [core/browser/widgets/SelectScripts.md](core/browser/widgets/SelectScripts.md): select_option page steps; has `inspectSelectScript`, `applyNativeSelectScript`, `collectOptionsScript`, `readSelectionScript`. - [core/browser/widgets/FieldScripts.md](core/browser/widgets/FieldScripts.md): set_date and set_slider page steps; has `inspectDateScript`, `inspectSliderScript`, `setValueScript`, `readValueScript`. - [core/browser/widgets/ListScripts.md](core/browser/widgets/ListScripts.md): collect_list page steps; has `collectListStepScript`, `scrollListScript`, `findLoadMoreScript`, `clickLoadMoreScript`. - [core/browser/BrowserController.md](core/browser/BrowserController.md): /api/browser REST controller over the tab manager; has tab, page, input, extraction and network-log actions. - [core/browser/BrowserMcpTools.md](core/browser/BrowserMcpTools.md): browser_* MCP tool controller derived from BrowserActions; has `TOOLS`, `ROUTES`, `createDirectHandler`. - [core/browser/controller/SelectorFallback.md](core/browser/controller/SelectorFallback.md): shared REST and MCP selector-plus-description fallback; has `run`. - [core/browser/controller/NavigationStamp.md](core/browser/controller/NavigationStamp.md): stamps urlChanged and newUrl on a reply; has `apply`. - [core/browser/controller/NetworkLogQuery.md](core/browser/controller/NetworkLogQuery.md): reads and filters a tab's request log; has `read`, `filter`. - [core/browser/controller/BrowserRequestParser.md](core/browser/controller/BrowserRequestParser.md): browser REST request shapes to tab manager arguments; has `tabId`, `updateAction`, `screenshotOptions`. - [core/browser/controller/RestFallbackReply.md](core/browser/controller/RestFallbackReply.md): fallback outcome to a REST reply; has `build`. - [core/browser/controller/FormFieldResolver.md](core/browser/controller/FormFieldResolver.md): re-resolves each field of a failed form fill; has `resolve`. - [core/browser/controller/PlainTab.md](core/browser/controller/PlainTab.md): tab to plain JSON; has `from`. - [core/browser/mcp/McpPayload.md](core/browser/mcp/McpPayload.md): result to MCP data payload; has `data`, `iso`. - [core/browser/mcp/ChildSelectorWarning.md](core/browser/mcp/ChildSelectorWarning.md): words child-selector defects for the model; has `compose`. - [core/browser/mcp/ScreenshotReply.md](core/browser/mcp/ScreenshotReply.md): browser_screenshot image reply; has `build`, `note`. - [core/browser/mcp/ExtractionReply.md](core/browser/mcp/ExtractionReply.md): browser_extract_data reply; has `build`. - [core/browser/mcp/SettledTab.md](core/browser/mcp/SettledTab.md): post-load state of a new tab; has `read`. - [core/browser/mcp/FallbackToolRunner.md](core/browser/mcp/FallbackToolRunner.md): runs a selector MCP tool with LLM and vision fallback; has `run`. - [core/browser/widgets/WidgetDriver.md](core/browser/widgets/WidgetDriver.md): widget primitives facade; has `selectOption`, `setDate`, `setSlider`, `collectList`. - [core/browser/widgets/WidgetPage.md](core/browser/widgets/WidgetPage.md): runs a widget page script with a uniform failure; has `run`. - [core/browser/widgets/DropdownOptions.md](core/browser/widgets/DropdownOptions.md): opens, lists, finds and clicks dropdown options; has `open`, `list`, `find`, `click`. - [core/browser/widgets/OptionSelector.md](core/browser/widgets/OptionSelector.md): select_option for native selects, listboxes and comboboxes; has `execute`. - [core/browser/widgets/DateSetter.md](core/browser/widgets/DateSetter.md): set_date for native and masked fields; has `execute`. - [core/browser/widgets/SliderSetter.md](core/browser/widgets/SliderSetter.md): set_slider for range inputs and ARIA sliders; has `execute`. - [core/browser/widgets/SliderKeyDriver.md](core/browser/widgets/SliderKeyDriver.md): moves an ARIA slider by trusted keys; has `execute`. - [core/browser/widgets/ListCollector.md](core/browser/widgets/ListCollector.md): collect_list over infinite, virtualized or load-more lists; has `execute`. - [core/browser/widgets/ListRowLedger.md](core/browser/widgets/ListRowLedger.md): rows seen once per key with virtualization detection; has `absorb`, `rows`. - [core/browser/TabViewManager.md](core/browser/TabViewManager.md): facade owning every tab view (lifecycle, strip, layout, focus, navigation, persistence); has `createTab`, `closeTab`, `switchToTab`, `navigate`, `goBack`, `goForward`, `setZoom`, `getAllTabs`, `findTabIdByWebContents`, `performAccelerator`. - [core/browser/tab-view/TabKinds.md](core/browser/tab-view/TabKinds.md): automation and internal tab kinds; has `isInternal`, `isAutomation`. - [core/browser/tab-view/NavigationGenerations.md](core/browser/tab-view/NavigationGenerations.md): drops load events from an older navigation; has `fresh`, `decide`, `normalizeUrl`. - [core/browser/tab-view/TabEntry.md](core/browser/tab-view/TabEntry.md): one tab's live state; has `applyNavEvent`, `isInternal`, `isInStrip`. - [core/browser/tab-view/TabRegistry.md](core/browser/tab-view/TabRegistry.md): tabs, strip order and active tab; has `add`, `remove`, `get`, `list`, `move`, `findIdByWebContents`. - [core/browser/tab-view/TabStateChannel.md](core/browser/tab-view/TabStateChannel.md): serializes tabs and sends them to the chrome of the window each is in; has `send`, `sendForTab`, `broadcast`, `serialize`. - [core/browser/tab-view/TabHosts.md](core/browser/tab-view/TabHosts.md): the windows tabs can live in (main plus detached), their bounds, and which window an IPC sender is; has `add`, `remove`, `windowOf`, `idForSender`. - [core/browser/tab-view/HiddenTabState.md](core/browser/tab-view/HiddenTabState.md): mutes hidden tabs and lowers their priority; has `sync`. - [core/browser/tab-view/TabErrorPage.md](core/browser/tab-view/TabErrorPage.md): the branded tab error and crash page; has `isErrorPage`, `targetOf`, `show`. - [core/browser/tab-view/TabViewFactory.md](core/browser/tab-view/TabViewFactory.md): builds and parks tab views, moves them between windows and attaches per-tab services; has `createView`, `attach`, `reparent`, `attachServices`. - [core/browser/tab-view/PersistedTabList.md](core/browser/tab-view/PersistedTabList.md): the persisted-tab list in settings; has `read`, `write`. - [core/browser/tab-view/SessionTabList.md](core/browser/tab-view/SessionTabList.md): the last session's strip snapshot in settings; has `read`, `write`. - [core/browser/tab-view/SessionTabs.md](core/browser/tab-view/SessionTabs.md): records the strip for "Restore tabs on restart"; has `saved`, `schedule`, `save`, `seal`. - [core/browser/tab-view/PersistedTabs.md](core/browser/tab-view/PersistedTabs.md): keep-alive tabs, reserved partitions, restore and revive; has `save`, `forget`, `list`, `restore`, `revive`. - [core/browser/tab-view/KeepAliveSweep.md](core/browser/tab-view/KeepAliveSweep.md): timed heartbeat, reload and revive of persisted tabs; has `start`, `stop`, `activateAll`. - [core/browser/tab-view/TabLayout.md](core/browser/tab-view/TabLayout.md): positions the active tab view; has `setBounds`, `applyTo`. - [core/browser/tab-view/TabActivation.md](core/browser/tab-view/TabActivation.md): switches, raises and focuses tabs; has `switchTo`, `cycle`, `selectByIndex`. - [core/browser/tab-view/TabLifecycle.md](core/browser/tab-view/TabLifecycle.md): creates, closes, moves and destroys tabs; has `create`, `close`, `reopenClosed`, `move`, `destroyAll`. - [core/browser/tab-view/TabTransfer.md](core/browser/tab-view/TabTransfer.md): moves tabs between windows and empties a closing detached window; has `move`, `evacuate`, `resync`. - [core/browser/tab-view/TabZoom.md](core/browser/tab-view/TabZoom.md): per-host tab zoom; has `set`, `get`, `by`. - [core/browser/tab-view/TabLoader.md](core/browser/tab-view/TabLoader.md): loads, reloads and stops tabs; has `navigate`, `waitForLoad`, `reload`, `stop`. - [core/browser/tab-view/TabHistory.md](core/browser/tab-view/TabHistory.md): tab session history; has `goBack`, `goForward`, `list`, `goToIndex`. - [core/browser/tab-view/TabPageTools.md](core/browser/tab-view/TabPageTools.md): find, print and DevTools; has `findInPage`, `print`, `toggleDevTools`. - [core/browser/tab-view/TabConsoleLog.md](core/browser/tab-view/TabConsoleLog.md): bounded per-tab console log; has `add`, `list`. - [core/browser/tab-view/TabLoadEvents.md](core/browser/tab-view/TabLoadEvents.md): keeps tab loading state in step with load events; has `wire`. - [core/browser/tab-view/TabPageEvents.md](core/browser/tab-view/TabPageEvents.md): crash recovery, shortcuts, find, favicon and console events; has `wire`. - [core/browser/tab-view/TabPopupHandler.md](core/browser/tab-view/TabPopupHandler.md): routes window.open to a popup or a new tab; has `wire`. - [core/browser/tab-view/TabPermissionHook.md](core/browser/tab-view/TabPermissionHook.md): applies the permission policy to tab sessions; has `install`. - [core/browser/tab-view/DataFileDownloadHook.md](core/browser/tab-view/DataFileDownloadHook.md): renders CSV and TSV inline and sends other downloads to the shelf; has `install`. - [core/browser/tab-view/TabAcceleratorActions.md](core/browser/tab-view/TabAcceleratorActions.md): main-handled browser shortcuts; has `perform`. - [core/browser/tab-view/TabNotificationRelay.md](core/browser/tab-view/TabNotificationRelay.md): relays intercepted web notifications; has `relay`. - [core/browser/tab-view/TabSessionCache.md](core/browser/tab-view/TabSessionCache.md): clears every tab session's cache; has `clearAll`. - [core/browser/tab-view/TabViewIpcController.md](core/browser/tab-view/TabViewIpcController.md): tab view IPC routes; has `register`. - [core/browser/ResolutionCache.md](core/browser/ResolutionCache.md): replays earlier LLM and vision element answers as validated selectors; has `isEnabled`, `replay`, `capture`, `remember`. - [core/browser/resolution-cache/ResolutionCacheKey.md](core/browser/resolution-cache/ResolutionCacheKey.md): resolution cache keys; has `of`, `urlPattern`, `description`, `kind`. - [core/browser/resolution-cache/ResolutionCacheSnapshot.md](core/browser/resolution-cache/ResolutionCacheSnapshot.md): persists the resolution cache in settings; has `load`, `save`. - [core/browser/resolution-cache/ResolutionCacheStore.md](core/browser/resolution-cache/ResolutionCacheStore.md): LRU resolution cache store with TTL; has `get`, `put`, `delete`, `clear`, `stats`, `flush`. - [core/browser/resolution-cache/TabPageAccess.md](core/browser/resolution-cache/TabPageAccess.md): tab URL and in-page script runs; has `urlOf`, `run`. - [core/browser/vision/VisualGroundingService.md](core/browser/vision/VisualGroundingService.md): finds described elements on screenshots with a vision model and clicks them; has `isAvailable`, `locate`, `locateInImage`, `clickDescribed`. - [core/browser/vision/GroundingCacheBridge.md](core/browser/vision/GroundingCacheBridge.md): visual grounding's resolution cache use; has `replay`, `noteHit`, `noteFailedClick`, `remember`. - [core/browser/TabManager.md](core/browser/TabManager.md): browser automation facade (page actions, refs, extraction, screenshots, evidence); has tab lifecycle, click, type, fill, scroll, observe, extract and widget methods. - [core/browser/tab-manager/TabPage.md](core/browser/tab-manager/TabPage.md): one tab for automation; has `run`, `runEnvelope`, `url`, `watchAction`. - [core/browser/tab-manager/NavigationReply.md](core/browser/tab-manager/NavigationReply.md): input-action reply with navigation details; has `build`. - [core/browser/tab-manager/TabControl.md](core/browser/tab-manager/TabControl.md): tab lifecycle and update actions; has `getAllTabs`, `createTab`, `getConsoleLogs`, `update`. - [core/browser/tab-manager/PageSource.md](core/browser/tab-manager/PageSource.md): page source by extraction type; has `read`. - [core/browser/tab-manager/PageSourceScripts.md](core/browser/tab-manager/PageSourceScripts.md): in-page source scripts per type; has `forType`. - [core/browser/tab-manager/PageAnalysisScript.md](core/browser/tab-manager/PageAnalysisScript.md): framework and selector-stability analysis script; has `SOURCE`. - [core/browser/tab-manager/ElementDigestFormatter.md](core/browser/tab-manager/ElementDigestFormatter.md): element digest to budgeted model text; has `format`. - [core/browser/tab-manager/PageObserver.md](core/browser/tab-manager/PageObserver.md): observe_page numbered element inventory; has `observe`. - [core/browser/tab-manager/TabScreenshot.md](core/browser/tab-manager/TabScreenshot.md): full-page or viewport screenshot with geometry and marks; has `capture`. - [core/browser/tab-manager/TargetScripts.md](core/browser/tab-manager/TargetScripts.md): target resolve, synthetic click and typing scripts; has `resolve`, `syntheticClick`, `typeValue`. - [core/browser/tab-manager/ElementClicker.md](core/browser/tab-manager/ElementClicker.md): trusted click with fallback and evidence; has `click`. - [core/browser/tab-manager/PointClicker.md](core/browser/tab-manager/PointClicker.md): click_at and point_info; has `clickAt`, `pointInfo`. - [core/browser/tab-manager/FieldTyper.md](core/browser/tab-manager/FieldTyper.md): focuses a field, sets its value, optionally submits; has `typeInto`. - [core/browser/tab-manager/KeyPresser.md](core/browser/tab-manager/KeyPresser.md): press_key with Enter emulation; has `pressKey`. - [core/browser/tab-manager/PageScroller.md](core/browser/tab-manager/PageScroller.md): scroll with settled position and evidence; has `scroll`. - [core/browser/tab-manager/FormFiller.md](core/browser/tab-manager/FormFiller.md): React-safe multi-field form fill; has `fill`. - [core/browser/tab-manager/ElementWaiter.md](core/browser/tab-manager/ElementWaiter.md): waits for a selector or its absence; has `waitFor`. - [core/browser/tab-manager/ElementInspector.md](core/browser/tab-manager/ElementInspector.md): get_element and batched selector checks; has `getElement`, `checkSelectors`. - [core/browser/tab-manager/TableReader.md](core/browser/tab-manager/TableReader.md): reads table or div-layout rows; has `read`. - [core/browser/tab-manager/DialogHandler.md](core/browser/tab-manager/DialogHandler.md): alert, confirm and prompt replacement; has `install`. - [core/browser/tab-manager/InteractableElements.md](core/browser/tab-manager/InteractableElements.md): interactable elements with stable selectors; has `list`. - [core/browser/tab-manager/AccessibleAttributeMatcher.md](core/browser/tab-manager/AccessibleAttributeMatcher.md): description to selector without an LLM; has `find`, `slug`. - [core/browser/tab-manager/ExtractDataScript.md](core/browser/tab-manager/ExtractDataScript.md): extract_data page script with child-selector diagnostics; has `build`. - [core/browser/tab-manager/RowDataExtractor.md](core/browser/tab-manager/RowDataExtractor.md): extract_data rows plus diagnostics; has `extract`. - [core/browser/tab-manager/SettleWaiter.md](core/browser/tab-manager/SettleWaiter.md): waits for the page to go quiet; has `wait`. - [core/browser/overlay/OverlayPreloadApi.md](core/browser/overlay/OverlayPreloadApi.md): `core/browser/overlay/OverlayPreloadApi.js` (entry: `core/browser/overlay/overlay-preload.js`). - [core/browser/ui/overlay/FaviconFallback.md](core/browser/ui/overlay/FaviconFallback.md): `FaviconFallback.apply(root)`: an `img.bookmark-favicon` that fails to load is replaced by `span.bookmark-favicon.bd-fallback` with its `data-letter` (or "?"). - [core/browser/ui/overlay/OverlayInput.md](core/browser/ui/overlay/OverlayInput.md): - `new OverlayInput(root, api, win)`, `attach()`: - mousedown on `[data-bd-action]` (prevented, so it lands before the URL bar's blur) -> `sendAction({ action, index, id, url })`; - mousemove over `[data-bd-index]` -> `sendHover({ index })`; - body enter/leave while a `.notification-log` is shown -> `sendHover({ hovering })`; - Ctrl/Cmd+C with a selection -> `execCommand('copy')`. - [core/browser/ui/overlay/OverlayUi.md](core/browser/ui/overlay/OverlayUi.md): `core/browser/ui/overlay/` (page: `core/browser/overlay/overlay.html`, loaded by ChromeOverlay). - [core/browser/ui/overlay/OverlayView.md](core/browser/ui/overlay/OverlayView.md): - `new OverlayView(root, api, win)`, `attach()`. - [core/browser/ui/error-page/ErrorPage.md](core/browser/ui/error-page/ErrorPage.md): `core/browser/ui/error-page/ErrorPage.js` (page: `core/browser/tab-view/error.html`, started by `entry.js`). - [core/browser/ui/error-page/NetErrorText.md](core/browser/ui/error-page/NetErrorText.md): the error page's wording per Chromium net error code. - [core/browser/tab-preload/ClickUpSocketScript.md](core/browser/tab-preload/ClickUpSocketScript.md): builds the page script that reports ClickUp's notifications from its websocket, the only page-side source for a site whose notifications only come through Web Push. - [core/browser/tab-preload/MainWorldScript.md](core/browser/tab-preload/MainWorldScript.md): builds the page script the tab preload runs in the page's MAIN world: the window.chrome and passkey shims, then the Notification and ServiceWorkerRegistration.showNotification overrides that report every notification with `window.postMessage`. - [core/browser/tab-preload/NotificationForwarder.md](core/browser/tab-preload/NotificationForwarder.md): the isolated-world half of notification interception: hears the main-world script's postMessage reports and sends each payload to the main process. - [core/browser/tab-preload/TabPreload.md](core/browser/tab-preload/TabPreload.md): installs the tab preload: injects the main-world script at document start and starts the notification forwarder. - [core/browser/ui/BrowserRenderer.md](core/browser/ui/BrowserRenderer.md): read-only tab view and tab events for extension renderers. - [core/browser/ui/TabSession.md](core/browser/ui/TabSession.md): renderer tab record with capped history. ## core/llm-server - [core/llm-server/GgufParser.md](core/llm-server/GgufParser.md): reads a GGUF header and, best-effort, its tensor layout; never throws; has `parseHeader`, `parse`. - [core/llm-server/GgufCursor.md](core/llm-server/GgufCursor.md): forward-only paged file reader with a byte cap and deadline; has `take`, `skip`, `skipStringArray`, `extendDeadline`. - [core/llm-server/GgufValueReader.md](core/llm-server/GgufValueReader.md): decodes GGUF primitive values from a cursor; has `uint32`, `uint64`, `length`, `string`, `scalar`, `array`. - [core/llm-server/GgufTensorLayout.md](core/llm-server/GgufTensorLayout.md): turns tensor entries into per-block KV, MTP and byte-placement facts; has `scan`, `merge`, `derive`. - [core/llm-server/GgufHeaderSummary.md](core/llm-server/GgufHeaderSummary.md): picks which GGUF metadata to keep and builds the header summary (role, parameter count, chat template, quant label); has `wants`, `numberOrNull`, `build`. - [core/llm-server/GgufRoleTable.md](core/llm-server/GgufRoleTable.md): the one table mapping a GGUF header (architecture, general.type, pooling, clip keys, tensor scan) to a role; has `headerRole`, `isKnownRole`. - [core/llm-server/HardwareSummary.md](core/llm-server/HardwareSummary.md): one-line hardware snapshot for fit-test results; has `summarize`, `shortGpuName`. - [core/llm-server/MemoryBandwidth.md](core/llm-server/MemoryBandwidth.md): GPU and RAM memory bandwidth with its source; has `tableGpuBandwidth`, `resolveGpuBandwidth`, `resolveRamBandwidth`. - [core/llm-server/LegacyChatMigration.md](core/llm-server/LegacyChatMigration.md): one-time non-destructive import of the old aiChat blob into ChatStore; has `run`, `readLegacyConversations`. - [core/llm-server/LocalApiIpcHandlers.md](core/llm-server/LocalApiIpcHandlers.md): IPC controller for the localhost OpenAI API toggle and port; has `register`. - [core/llm-server/eval/Aggregator.md](core/llm-server/eval/Aggregator.md): per-variant eval means, pass rates and A/B diffs; has `aggregate`, `diffVariants`, `formatDiff`. - [core/llm-server/eval/GoldenSet.md](core/llm-server/eval/GoldenSet.md): loads and checks golden eval tasks; has `loadGoldenTasks`, `validateTask`. - [core/llm-server/eval/Scorer.md](core/llm-server/eval/Scorer.md): scores a task's transcripts to 0..1 with a per-check breakdown; has `scoreTask`, `normalizeTurns`, `harvestHealth`. - [core/llm-server/eval/TurnChecks.md](core/llm-server/eval/TurnChecks.md): builds the weighted checks for one turn's expect block; has `collect`. - [core/llm-server/eval/ExpectationMatcher.md](core/llm-server/eval/ExpectationMatcher.md): JSON value, param and tool-call matchers, shared with trigger gating; has `matchValue`, `matchParams`, `callMatches`, `safeRegex`. - [core/llm-server/eval/TranscriptHealth.md](core/llm-server/eval/TranscriptHealth.md): rolls up health signals from transcripts; has `harvest`. - [core/llm-server/gambit/ExecutionCheck.md](core/llm-server/gambit/ExecutionCheck.md): runs the model's JS for a gambit task and grades each case; has `runExecutionCheck`, `extractCode`, `checkCase`. - [core/llm-server/gambit/CodeSandbox.md](core/llm-server/gambit/CodeSandbox.md): runs code once in a vm with sync and async timeouts; has `runOnce`. - [core/llm-server/gambit/LiveArtifactStubs.md](core/llm-server/gambit/LiveArtifactStubs.md): stand-ins for live-artifact globals during checks; has `create`. - [core/llm-server/models/CuratedModelCatalog.md](core/llm-server/models/CuratedModelCatalog.md): curated GGUF picks for onboarding with download URLs; has `resolveUrl`, `listCatalog`. - [core/llm-server/models/HfHubClient.md](core/llm-server/models/HfHubClient.md): HF Hub transport, revision pinning and repo id helpers, plus the full hfHub.js mapping; has `get`, `getRepoFile`, `resolveRevision`, `resolveUrl`, `fileUrl`, `treeUrl`, `normalizeRepoId`. - [core/llm-server/models/HfPinnedRepo.md](core/llm-server/models/HfPinnedRepo.md): one pull's view of a Hub repo pinned to a commit sha with each file's size and sha256; has `pin`, `fromTree`, `urlFor`, `fileFor`. - [core/llm-server/models/HfModelSearch.md](core/llm-server/models/HfModelSearch.md): GGUF and MLX repo search on the Hub; has `search`. - [core/llm-server/models/HfGgufRepo.md](core/llm-server/models/HfGgufRepo.md): a GGUF repo's header, quant variants and companions, URLs pinned to one commit with digests; has `fetchVariants`, `fetchCompanions`, `fetchMmprojs`. - [core/llm-server/models/HfMlxRepo.md](core/llm-server/models/HfMlxRepo.md): describes an MLX repo for snapshot download, pinned to one commit with file digests; has `fetchInfo`, `quantFromName`. - [core/llm-server/models/HfReadme.md](core/llm-server/models/HfReadme.md): fetches a repo's model card, frontmatter stripped; has `fetch`. - [core/llm-server/models/GgufFileName.md](core/llm-server/models/GgufFileName.md): reads quant, shard and companion type from GGUF file names; has `quantOf`, `split`, `shardIndex`, `isMtpHead`, `isMmproj`. - [core/llm-server/models/GgufVariantGrouper.md](core/llm-server/models/GgufVariantGrouper.md): turns a repo tree into one variant per quant; has `group`. - [core/llm-server/models/GgufCompanionPicker.md](core/llm-server/models/GgufCompanionPicker.md): picks the vision projector and MTP head to download; has `bestMmproj`, `bestMtp`. - [core/llm-server/models/ModelCatalogRegistry.md](core/llm-server/models/ModelCatalogRegistry.md): add-on models contributed to LLM Setup by extensions; has `shared` plus registry methods. - [core/llm-server/models/MoeEstimator.md](core/llm-server/models/MoeEstimator.md): splits MoE weight bytes into expert, GPU-resident and host parts; has `estimateSplit`. - [core/llm-server/router/RouterPrompt.md](core/llm-server/router/RouterPrompt.md): prompt, grammar and parser for the tool-group router; has `buildRouterPrompt`, `parseRouterAnswer`. - [core/llm-server/runtimes/RuntimeCatalogRegistry.md](core/llm-server/runtimes/RuntimeCatalogRegistry.md): extension-contributed inference runtimes and launch hooks; has `shared`, `register`, `hooksFor`, `runtimesForModelKind`. - [core/llm-server/server/FailureInterpreter.md](core/llm-server/server/FailureInterpreter.md): turns a dead runtime's stderr into a diagnosis; has `interpret`, `extractUnknownFlag`. - [core/llm-server/server/ModelCapsCache.md](core/llm-server/server/ModelCapsCache.md): remembers each model file's reasoning capabilities; has `remember`, `recall`, `recallProbe`, `modelKeyOf`. - [core/llm-server/server/ModelFamilies.md](core/llm-server/server/ModelFamilies.md): detects a model's tuning family and its launch args and switches; has `detect`, `launchArgs`, `speculativeArgs`, `nativeToolCalls`, `kvQuantUnsafe`, `ngramSpec`. - [core/llm-server/server/ModelFamilyTable.md](core/llm-server/server/ModelFamilyTable.md): data-only table of each family's sampler, flags and drafters; has `FAMILIES`, `SAMPLER_CLI`. - [core/llm-server/server/ThinkingProbe.md](core/llm-server/server/ThinkingProbe.md): learns a chat template's reasoning controls; has `probe`, `templateHashOf`. - [core/llm-server/server/ThinkingFacts.md](core/llm-server/server/ThinkingFacts.md): the single shape for reasoning facts; has `derive`, `fromRegex`, `offersControl`. - [core/llm-server/server/ThinkingFactsDeriver.md](core/llm-server/server/ThinkingFactsDeriver.md): derives reasoning facts from template renders; has `execute`. - [core/llm-server/server/ThinkingProbePlan.md](core/llm-server/server/ThinkingProbePlan.md): builds the bounded render plan and request bodies; has `buildShapes`, `buildPlan`, `requestBody`. - [core/llm-server/server/ThinkingVocabulary.md](core/llm-server/server/ThinkingVocabulary.md): data-only effort vocabulary and think markers. - [core/llm-server/server/RpcPeers.md](core/llm-server/server/RpcPeers.md): borrows peer GPUs for a launch and heartbeats the leases; has `acquireForLaunch`, `releaseAll`, `isActive`, `activePeers`. - [core/llm-server/server/UnsupportedFlagMemory.md](core/llm-server/server/UnsupportedFlagMemory.md): remembers launch flags a runtime build rejected; has `runtimeKey`, `learnedFlags`, `withLearnedFlags`, `remember`. - [core/llm-server/server/UserArgs.md](core/llm-server/server/UserArgs.md): shell-style split of the user's extra llama.cpp flags; has `splitShellArgs`, `normalize`. - [core/llm-server/server/chat/ChatAdapter.md](core/llm-server/server/chat/ChatAdapter.md): base class for inference-server wire-protocol adapters; has `authHeaders`, `healthCheck`, `chat`, `protocolId`. - [core/llm-server/validation/ICodeLanguageValidator.md](core/llm-server/validation/ICodeLanguageValidator.md): base class for per-language code validators; has `name`, `languages`, `validate`. - [core/llm-server/ChatStore.md](core/llm-server/ChatStore.md): repository for chat conversations, messages, variants and mode meta; has conversation CRUD, `restoreConversationTimestamps`, message CRUD, variants. - [core/llm-server/SystemDiagnostics.md](core/llm-server/SystemDiagnostics.md): facade over the hardware report and the Windows GPU fixes; has `gather`, `addDirectoryToUserPath`, `recoverDisplayDevice`, `setPcieAspmOff`. - [core/llm-server/diagnostics/DiagnosticsGatherer.md](core/llm-server/diagnostics/DiagnosticsGatherer.md): builds the full fail-soft diagnostics report, NPUs, shared GPU memory and Linux llama.cpp device VRAM included; has `gather`. - [core/llm-server/diagnostics/HostProbe.md](core/llm-server/diagnostics/HostProbe.md): platform, RAM totals and CPU; has `platform`, `memory`, `cpu`. - [core/llm-server/diagnostics/RamModuleProbe.md](core/llm-server/diagnostics/RamModuleProbe.md): per-DIMM RAM details on each platform; has `probe`. - [core/llm-server/diagnostics/RamModuleTextParser.md](core/llm-server/diagnostics/RamModuleTextParser.md): reads system_profiler and dmidecode text; has `parseSystemProfiler`, `parseDmidecode`. - [core/llm-server/diagnostics/WindowsRamModuleParser.md](core/llm-server/diagnostics/WindowsRamModuleParser.md): maps Win32_PhysicalMemory rows; has `parseRows`. - [core/llm-server/diagnostics/CudaProbe.md](core/llm-server/diagnostics/CudaProbe.md): finds nvidia-smi and reads CUDA; has `probe`. - [core/llm-server/diagnostics/CudaDeviceParser.md](core/llm-server/diagnostics/CudaDeviceParser.md): nvidia-smi CSV and XML parsing; has `parseDevices`, `parsePciDeviceId`, `parseCudaVersion`. - [core/llm-server/diagnostics/GpuProbe.md](core/llm-server/diagnostics/GpuProbe.md): GPU adapters with VRAM (nvidia-smi, Windows registry or Linux sysfs) and PCIe details; has `probe`. - [core/llm-server/diagnostics/GpuAdapterMatcher.md](core/llm-server/diagnostics/GpuAdapterMatcher.md): pairs and de-duplicates GPU adapters across sources; has `matchNvidiaByPci`, `matchNvidiaByName`, `matchRegistry`, `dedupe`. - [core/llm-server/diagnostics/RegistryVramProbe.md](core/llm-server/diagnostics/RegistryVramProbe.md): Windows 64-bit VRAM from the registry; has `probe`. - [core/llm-server/diagnostics/LinuxDrmVramProbe.md](core/llm-server/diagnostics/LinuxDrmVramProbe.md): Linux GPU ids, VRAM and iGPU placement from sysfs; has `probe`, `isIntegrated`. - [core/llm-server/diagnostics/UnifiedMemoryProbe.md](core/llm-server/diagnostics/UnifiedMemoryProbe.md): the Apple Silicon GPU's unified-memory working set as VRAM; has `applies`, `probe`, `workingSet`. - [core/llm-server/diagnostics/SharedGpuMemory.md](core/llm-server/diagnostics/SharedGpuMemory.md): VRAM figures for unified-memory and integrated GPUs from system RAM; has `apply`, `isIntegrated`. - [core/llm-server/diagnostics/LlamaDeviceProbe.md](core/llm-server/diagnostics/LlamaDeviceProbe.md): Linux GPU memory from an installed llama.cpp build's `--list-devices`; has `probe`, `parse`. - [core/llm-server/diagnostics/LlamaDeviceVram.md](core/llm-server/diagnostics/LlamaDeviceVram.md): folds the llama.cpp device list into Intel and AMD adapters missing VRAM; has `apply`, `vendorOf`, `pair`. - [core/llm-server/diagnostics/NpuProbe.md](core/llm-server/diagnostics/NpuProbe.md): Intel and other NPUs from Windows PnP or Linux accel devices, with whether OpenVINO can open them; has `probe`, `hasIntelNpu`, `usableIntelNpus`. - [core/llm-server/diagnostics/GpuHealthProbe.md](core/llm-server/diagnostics/GpuHealthProbe.md): Windows Plug-and-Play health of display adapters; has `probe`, `classify`. - [core/llm-server/diagnostics/PnpDisplayDevices.md](core/llm-server/diagnostics/PnpDisplayDevices.md): lists display-class Plug-and-Play devices; has `list`, `isNonGpu`. - [core/llm-server/diagnostics/PcieAspm.md](core/llm-server/diagnostics/PcieAspm.md): reads and turns off PCIe link power saving; has `read`, `setOff`, `parse`. - [core/llm-server/diagnostics/DisplayDeviceRecovery.md](core/llm-server/diagnostics/DisplayDeviceRecovery.md): elevated recovery of a faulted display adapter; has `recover`, `interpret`. - [core/llm-server/diagnostics/DisplayRecoveryScripts.md](core/llm-server/diagnostics/DisplayRecoveryScripts.md): the display recovery PowerShell scripts; has `elevated`, `launcher`. - [core/llm-server/diagnostics/UserPathEditor.md](core/llm-server/diagnostics/UserPathEditor.md): adds the nvidia-smi folder to the user PATH; has `addDirectory`. - [core/llm-server/diagnostics/UserPathScripts.md](core/llm-server/diagnostics/UserPathScripts.md): the PATH PowerShell scripts; has `displayCommand`, `persistScript`. - [core/llm-server/diagnostics/DiskProbe.md](core/llm-server/diagnostics/DiskProbe.md): volumes and free space with fallbacks; has `probe`. - [core/llm-server/diagnostics/DiskTableParser.md](core/llm-server/diagnostics/DiskTableParser.md): PowerShell, wmic and df disk tables; has `fromLogicalDiskRows`, `parseWmicCsv`, `parseDf`. - [core/llm-server/diagnostics/ResourceBudget.md](core/llm-server/diagnostics/ResourceBudget.md): RAM and VRAM budget after reserves; has `compute`. - [core/llm-server/diagnostics/DiagnosticsCommand.md](core/llm-server/diagnostics/DiagnosticsCommand.md): runs a probe command that never throws; has `run`, `isBinaryMissing`. - [core/llm-server/diagnostics/PowerShellRunner.md](core/llm-server/diagnostics/PowerShellRunner.md): runs a script on the first PowerShell found; has `run`, `quote`, `failureMessage`. - [core/llm-server/diagnostics/PowerShellJson.md](core/llm-server/diagnostics/PowerShellJson.md): reads ConvertTo-Json output as rows; has `parseRows`, `stripBom`. - [core/llm-server/diagnostics/HardwareText.md](core/llm-server/diagnostics/HardwareText.md): placeholder-aware hardware text and numbers; has `clean`, `numberOrNull`. - [core/llm-server/eval/EvalRunner.md](core/llm-server/eval/EvalRunner.md): runs eval tasks across variants and models; has `runEval`, `diffFromRuns`, `run`. - [core/llm-server/gambit/GambitReport.md](core/llm-server/gambit/GambitReport.md): model compatibility score with health, coverage and worst decile; has `build`, `band`, `orderGroups`. - [core/llm-server/gambit/GambitHealth.md](core/llm-server/gambit/GambitHealth.md): gambit health totals and clean-turn score; has `totals`, `score`. - [core/llm-server/gambit/GambitReportFormatter.md](core/llm-server/gambit/GambitReportFormatter.md): one-screen console summary of a gambit report; has `format`. - [core/llm-server/GgufHeaderParseWorker.md](core/llm-server/GgufHeaderParseWorker.md): parses GGUF headers on a worker thread (entry ggufParseWorker.js); has `listen`, `handle`. - [core/llm-server/Preflight.md](core/llm-server/Preflight.md): boot-time list of configuration problems for the LLM tab; has `collectIssues`, `collect`. - [core/llm-server/PreflightRuntimeIssue.md](core/llm-server/PreflightRuntimeIssue.md): unknown, missing or not-ready runtime issue; has `findRuntime`, `forRow`. - [core/llm-server/PreflightSysdepsIssue.md](core/llm-server/PreflightSysdepsIssue.md): Linux missing-system-libraries issue; has `collect`. - [core/llm-server/models/ModelDownload.md](core/llm-server/models/ModelDownload.md): resumable single-file model download with pause and cancel; has `start`. - [core/llm-server/models/MlxRepoDownload.md](core/llm-server/models/MlxRepoDownload.md): downloads an MLX repo snapshot at its pinned commit with aggregate progress, safe paths and a free-space guard; has `start`. - [core/llm-server/models/HwBudget.md](core/llm-server/models/HwBudget.md): turns diagnostics into the flat hardware budget the pickers read; has `build`. - [core/llm-server/models/ModelName.md](core/llm-server/models/ModelName.md): stable model keys and display names; has `key`, `prettify`, `resolveDisplayName`. - [core/llm-server/models/ModelNameToken.md](core/llm-server/models/ModelNameToken.md): noise and styling rules for one model-name token; has `isNoise`, `isQuantOrFormat`, `style`. - [core/llm-server/models/libraries/ExistingLibraryScanner.md](core/llm-server/models/libraries/ExistingLibraryScanner.md): finds GGUFs in LM Studio, the HF cache and Ollama; has `scan`. - [core/llm-server/models/libraries/ExistingModelImporter.md](core/llm-server/models/libraries/ExistingModelImporter.md): adopts a found model by link, never copy; has `importModel`. - [core/llm-server/models/libraries/LibrarySource.md](core/llm-server/models/libraries/LibrarySource.md): base class for another tool's model library; has `id`, `label`, `roots`, `scan`. - [core/llm-server/models/libraries/LmStudioSource.md](core/llm-server/models/libraries/LmStudioSource.md): LM Studio library source; has `roots`, `scan`. - [core/llm-server/models/libraries/HfCacheSource.md](core/llm-server/models/libraries/HfCacheSource.md): Hugging Face hub cache source; has `roots`, `scan`. - [core/llm-server/models/libraries/OllamaSource.md](core/llm-server/models/libraries/OllamaSource.md): Ollama manifest and blob source; has `roots`, `scan`. - [core/llm-server/models/libraries/GgufTreeWalker.md](core/llm-server/models/libraries/GgufTreeWalker.md): depth-bounded walk for adoptable GGUFs; has `walk`, `isAdoptableName`. - [core/llm-server/models/libraries/FileProbe.md](core/llm-server/models/libraries/FileProbe.md): never-throwing filesystem checks; has `stat`, `isDirectory`, `readDirectory`, `realPath`. - [core/llm-server/runtimes/LlmRuntimeCatalog.md](core/llm-server/runtimes/LlmRuntimeCatalog.md): LLM runtime catalog with extension runtimes merged at read time; has `shared`, `getCatalog`, `getById`, `fingerprint`, `cudaRuntimePreference`. - [core/llm-server/runtimes/LlmRuntimeDeclarations.md](core/llm-server/runtimes/LlmRuntimeDeclarations.md): data-only table of the llama.cpp (CPU/Metal, CUDA, Vulkan, ROCm, SYCL, OpenVINO), MLX and Harmony runtimes; has `RUNTIMES`. - [core/llm-server/runtimes/OpenVinoDevice.md](core/llm-server/runtimes/OpenVinoDevice.md): the OpenVINO runtime's compute device (GPU, NPU or CPU) and its spawn variable; has `normalize`, `resolve`, `appliesTo`, `spawnEnv`. - [core/llm-server/server/chat/RequestProfile.md](core/llm-server/server/chat/RequestProfile.md): per-runtime chat body translation and effort rounding; has `apply`, `mapEffortLevel`, `fromThinking`. - [core/llm-server/server/ThinkingKnobs.md](core/llm-server/server/ThinkingKnobs.md): turns a client's thinking spellings into llama-server knobs; has `extra`, `apply`, `refusal`. - [core/llm-server/server/ModelIdRewriter.md](core/llm-server/server/ModelIdRewriter.md): stream transform that rewrites the model field in JSON and SSE; has `create`. - [core/llm-server/server/UpstreamProxy.md](core/llm-server/server/UpstreamProxy.md): forwards one JSON POST to llama-server and streams the reply; has `forward`. - [core/llm-server/server/OpenAiModelList.md](core/llm-server/server/OpenAiModelList.md): builds the /v1/models replies; has `build`, `single`. - [core/llm-server/server/OpenAiLocalRouter.md](core/llm-server/server/OpenAiLocalRouter.md): route controller for the localhost OpenAI API; has `create`. - [core/llm-server/server/ByteLadder.md](core/llm-server/server/ByteLadder.md): main-process byte formatter; has `format`. - [core/llm-server/server/MlxLaunchPlanner.md](core/llm-server/server/MlxLaunchPlanner.md): plans an mlx_lm.server launch; has `plan`. - [core/llm-server/server/decode/DecodeFormula.md](core/llm-server/server/decode/DecodeFormula.md): the memory-bound decode speed formula; has `estimateTps`, `tpsFromMs`, `describeTps`. - [core/llm-server/server/decode/GpuLayerAssignment.md](core/llm-server/server/decode/GpuLayerAssignment.md): maps offloaded blocks to GPUs; has `assign`. - [core/llm-server/server/decode/PlanDecodeEstimator.md](core/llm-server/server/decode/PlanDecodeEstimator.md): decode speed prediction for a resolved launch plan; has `estimate`. - [core/llm-server/server/decode/CatalogDecodeEstimator.md](core/llm-server/server/decode/CatalogDecodeEstimator.md): pre-download decode speed guess for a catalog row; has `estimate`. - [core/llm-server/validation/validators/EmbeddedScriptCheck.md](core/llm-server/validation/validators/EmbeddedScriptCheck.md): compiles page scripts carried in template literals (handed to executeJs, or IIFE-shaped), which ESLint never parses; has `check`. - [core/llm-server/validation/validators/JavaScriptValidator.md](core/llm-server/validation/validators/JavaScriptValidator.md): ESLint validator for JS and JSX; has `validate`. - [core/llm-server/validation/validators/TypeScriptValidator.md](core/llm-server/validation/validators/TypeScriptValidator.md): TypeScript compiler validator for TS and TSX; has `validate`. - [core/llm-server/validation/validators/JsonValidator.md](core/llm-server/validation/validators/JsonValidator.md): JSON and JSONC validator; has `validate`, `locateError`. - [core/llm-server/validation/validators/JsonCommentStripper.md](core/llm-server/validation/validators/JsonCommentStripper.md): string-aware JSONC comment remover; has `strip`. - [core/llm-server/validation/validators/PowerShellValidator.md](core/llm-server/validation/validators/PowerShellValidator.md): syntax-only PowerShell validator in wasm; has `validate`. - [core/llm-server/validation/validators/PowerShellGrammar.md](core/llm-server/validation/validators/PowerShellGrammar.md): cached tree-sitter PowerShell grammar loader; has `load`, `createParser`. - [core/llm-server/validation/validators/PowerShellSyntaxCollector.md](core/llm-server/validation/validators/PowerShellSyntaxCollector.md): turns parse errors into diagnostics; has `collect`. - [core/llm-server/chat/ArtifactStore.md](core/llm-server/chat/ArtifactStore.md): persists artifact version chains and writes each version to a standalone HTML file; has `create`, `createVersion`, `get`, `renderedHtml`, `list`, `versions`, `delete`, `deleteRoot`, `urlFor`. - [core/llm-server/chat/artifacts/ArtifactContent.md](core/llm-server/chat/artifacts/ArtifactContent.md): artifact kinds and how each kind's content is stored; has `typeOf`, `isMedia`, `normalize`. - [core/llm-server/chat/artifacts/ArtifactDocument.md](core/llm-server/chat/artifacts/ArtifactDocument.md): wraps artifact content into a self-contained styled document; has `render`. - [core/llm-server/chat/artifacts/ArtifactDocumentShell.md](core/llm-server/chat/artifacts/ArtifactDocumentShell.md): the dark artifact document chrome; has `wrap`. - [core/llm-server/chat/artifacts/ArtifactMarkdown.md](core/llm-server/chat/artifacts/ArtifactMarkdown.md): compact markdown converter for markdown artifacts; has `toHtml`. - [core/llm-server/chat/artifacts/LiveModuleDocument.md](core/llm-server/chat/artifacts/LiveModuleDocument.md): standalone live-module page with store and luma wiring; has `renderBody`. - [core/llm-server/chat/artifacts/HtmlText.md](core/llm-server/chat/artifacts/HtmlText.md): HTML escaping for artifact documents; has `escape`. - [core/llm-server/chat/ArtifactDataStore.md](core/llm-server/chat/ArtifactDataStore.md): per-chain key/value data for live artifacts; has `all`, `get`, `mutate`, `set`, `remove`, `deleteForRoot`. - [core/llm-server/chat/artifacts/ArtifactDataQuota.md](core/llm-server/chat/artifacts/ArtifactDataQuota.md): size limits on live-artifact data; has `checkEntry`, `checkObject`. - [core/llm-server/chat/IntervalTaskStore.md](core/llm-server/chat/IntervalTaskStore.md): base class for recurring-task stores with run history; has `create`, `get`, `list`, `due`, `update`, `recordCompletion`, `delete`, `recordRunStart`, `recordRunFinish`, `listRuns`. - [core/llm-server/chat/ArtifactTaskStore.md](core/llm-server/chat/ArtifactTaskStore.md): scheduled artifact-refresh tasks; has `listByRoot`, `deleteByRoot` plus the base methods. - [core/llm-server/chat/ScheduledTaskStore.md](core/llm-server/chat/ScheduledTaskStore.md): chat scheduled tasks; has `listWithRunCounts`, `listByConversation`, `getByConversation` plus the base methods. - [core/llm-server/chat/ChatDataWipe.md](core/llm-server/chat/ChatDataWipe.md): clears all conversations and artifacts from the database; has `execute`. - [core/llm-server/chat/MessageTree.md](core/llm-server/chat/MessageTree.md): the branch structure of a conversation's messages (parent links, variants as siblings) and the active path the thread shows; also the parent links backfilled for pre-branching rows. - [core/llm-server/chat/JsonOutliner.md](core/llm-server/chat/JsonOutliner.md): budgeted JSONPath outline of an oversized tool result, with inferred schema and statistics; has `outline`. - [core/llm-server/chat/outline/BoundedJson.md](core/llm-server/chat/outline/BoundedJson.md): compact JSON that gives up past a character cap; has `stringify`. - [core/llm-server/chat/outline/ValueKind.md](core/llm-server/chat/outline/ValueKind.md): the JSON kind of a value; has `of`, `normalize`, `isPresent`. - [core/llm-server/chat/outline/JsonPath.md](core/llm-server/chat/outline/JsonPath.md): JSONPath strings for outline labels; has `child`, `everyElement`. - [core/llm-server/chat/outline/ElementSampler.md](core/llm-server/chat/outline/ElementSampler.md): evenly spread array indices to profile; has `indices`. - [core/llm-server/chat/outline/StringPreview.md](core/llm-server/chat/outline/StringPreview.md): a short quoted prefix of a string; has `quote`. - [core/llm-server/chat/outline/CountPhrase.md](core/llm-server/chat/outline/CountPhrase.md): a count with its singular or plural noun; has `of`, `noun`. - [core/llm-server/chat/outline/ValueProfile.md](core/llm-server/chat/outline/ValueProfile.md): kinds, statistics, fields and elements observed at one path; has `record`, `field`, `elementProfile`. - [core/llm-server/chat/outline/ProfileWalker.md](core/llm-server/chat/outline/ProfileWalker.md): walks a value into a bounded profile tree; has `profile`. - [core/llm-server/chat/outline/ProfileSummary.md](core/llm-server/chat/outline/ProfileSummary.md): the words for one profiled path; has `describe`. - [core/llm-server/chat/outline/OutlineNode.md](core/llm-server/chat/outline/OutlineNode.md): one outline line and its children; has `line`, `moreLine`. - [core/llm-server/chat/outline/OutlineTreeBuilder.md](core/llm-server/chat/outline/OutlineTreeBuilder.md): profile tree to outline nodes with optional-field marks; has `build`. - [core/llm-server/chat/outline/OutlineBudget.md](core/llm-server/chat/outline/OutlineBudget.md): breadth-first line admission within a token budget; has `select`, `cost`. - [core/llm-server/chat/outline/OutlineRenderer.md](core/llm-server/chat/outline/OutlineRenderer.md): writes admitted lines in document order; has `render`. - [core/llm-server/chat/outline/TokenSizeLabel.md](core/llm-server/chat/outline/TokenSizeLabel.md): approximate token counts for the model ("4,300"); has `format`, `ofText`. - [core/llm-server/chat/outline/stats/ValueStats.md](core/llm-server/chat/outline/stats/ValueStats.md): base for per-kind path statistics; has `observe`, `describe`, `detail`. - [core/llm-server/chat/outline/stats/NumberStats.md](core/llm-server/chat/outline/stats/NumberStats.md): integer/number range, distinct count, codes; extends ValueStats. - [core/llm-server/chat/outline/stats/StringStats.md](core/llm-server/chat/outline/stats/StringStats.md): string lengths, distinct count, categories, example; extends ValueStats. - [core/llm-server/chat/outline/stats/BooleanStats.md](core/llm-server/chat/outline/stats/BooleanStats.md): true and false counts; extends ValueStats. - [core/llm-server/chat/outline/stats/RangeTracker.md](core/llm-server/chat/outline/stats/RangeTracker.md): min and max phrased as a range; has `observe`, `phrase`, `format`. - [core/llm-server/chat/outline/stats/DistinctCounter.md](core/llm-server/chat/outline/stats/DistinctCounter.md): capped distinct counting and enumerations; has `add`, `label`, `enumeration`. - [core/llm-server/chat/LlmTrace.md](core/llm-server/chat/LlmTrace.md): one JSON line per model call per conversation, off by default; has `enabled`, `status`, `instrument`, `record`, `deleteFor`, `wipeAll`. - [core/llm-server/chat/trace/LlmTraceWriter.md](core/llm-server/chat/trace/LlmTraceWriter.md): non-blocking trace appender with rotation; has `append`, `fileFor`, `deleteFor`, `wipeAll`. - [core/llm-server/chat/trace/LlmTraceRecord.md](core/llm-server/chat/trace/LlmTraceRecord.md): turns a trace record into a bounded, image-free line; has `toLine`, `sanitizeBody`. - [core/llm-server/chat/trace/LlmCallRecorder.md](core/llm-server/chat/trace/LlmCallRecorder.md): wraps dispatch hooks and records each call once; has `wrap`. - [core/llm-server/chat/ToolPresentation.md](core/llm-server/chat/ToolPresentation.md): replay-safe tool-card metadata kept away from the model; has `present`, `metaFromEntry`, `withDiffBasis`. - [core/llm-server/chat/presentation/ToolCardPresenter.md](core/llm-server/chat/presentation/ToolCardPresenter.md): base class for tool card presenters; has `present`, `validate`. - [core/llm-server/chat/presentation/DiffCardPresenter.md](core/llm-server/chat/presentation/DiffCardPresenter.md): diff card for file-editing tools; has `attachBasis`, `present`, `validate`. - [core/llm-server/chat/presentation/ReadCardPresenter.md](core/llm-server/chat/presentation/ReadCardPresenter.md): read-slice card for read_file; has `present`, `validate`. - [core/llm-server/chat/ToolResultSpill.md](core/llm-server/chat/ToolResultSpill.md): writes over-budget tool results to disk; has `createWriter`, `deleteFor`, `wipeAll`, `setBaseDir`. - [core/llm-server/chat/spill/SpillWriter.md](core/llm-server/chat/spill/SpillWriter.md): one run's spill files; has `write`, `dir`, `readable`. - [core/llm-server/chat/spill/GitExclude.md](core/llm-server/chat/spill/GitExclude.md): adds a pattern to .git/info/exclude once; has `addOnce`. - [core/llm-server/chat/AppOwnedDir.md](core/llm-server/chat/AppOwnedDir.md): app-owned subfolder of the managed base dir; has `resolve`. - [core/llm-server/chat/FileNameSegment.md](core/llm-server/chat/FileNameSegment.md): turns an id into one safe path segment; has `from`. - [core/llm-server/chat/WorkspaceFiles.md](core/llm-server/chat/WorkspaceFiles.md): root-jailed file API behind the Code view; has `resolveRoot`, `listDir`, `readFile`, `writeFile`, `createEntry`, `renameEntry`, `removeEntry`. - [core/llm-server/chat/workspace/WorkspaceRootResolver.md](core/llm-server/chat/workspace/WorkspaceRootResolver.md): finds a conversation's workspace folder; has `resolve`. - [core/llm-server/chat/workspace/DirectoryListing.md](core/llm-server/chat/workspace/DirectoryListing.md): one capped, dirs-first directory level; has `list`. - [core/llm-server/chat/workspace/EditorFileReader.md](core/llm-server/chat/workspace/EditorFileReader.md): reads a file for the editor, refusing binary and oversized files; has `read`. - [core/llm-server/chat/workspace/WorkspaceFileIpcHandlers.md](core/llm-server/chat/workspace/WorkspaceFileIpcHandlers.md): IPC controller for workspace files; has `register`. - [core/llm-server/chat/triggers/TriggerGating.md](core/llm-server/chat/triggers/TriggerGating.md): trigger cooldown and batch gating; has `normalize`, `buildBatchEvent`, `describe`. - [core/llm-server/chat/triggers/TriggerFilter.md](core/llm-server/chat/triggers/TriggerFilter.md): dotted-path filter rules over an event; has `getPath`, `normalize`, `evaluate`. - [core/llm-server/chat/triggers/FileWatchSource.md](core/llm-server/chat/triggers/FileWatchSource.md): file and folder trigger source; has `reconcile`, `validateDir`, `status`, `watchFor`, `stopAll`. - [core/llm-server/chat/triggers/file/FileWatch.md](core/llm-server/chat/triggers/file/FileWatch.md): one trigger's file watch with settle, snapshot and catch-up; has `start`, `stop`. - [core/llm-server/chat/triggers/file/WatchFolder.md](core/llm-server/chat/triggers/file/WatchFolder.md): watch-folder safety rules; has `validate`, `resolveInside`. - [core/llm-server/chat/triggers/file/FileGlobMatcher.md](core/llm-server/chat/triggers/file/FileGlobMatcher.md): glob matcher with scratch-file ignores; has `create`. - [core/llm-server/chat/triggers/file/FileEventBuilder.md](core/llm-server/chat/triggers/file/FileEventBuilder.md): builds sanitised file events with previews; has `build`, `forPath`. - [core/llm-server/chat/triggers/file/FolderScanner.md](core/llm-server/chat/triggers/file/FolderScanner.md): bounded folder snapshot; has `scan`. - [core/llm-server/chat/triggers/file/CatchUpPlan.md](core/llm-server/chat/triggers/file/CatchUpPlan.md): what to replay after a restart; has `build`, `sameSignature`. - [core/llm-server/server/AnthropicMessagesRouter.md](core/llm-server/server/AnthropicMessagesRouter.md): Express controller for the local Anthropic Messages API; has `create`. - [core/llm-server/server/anthropic/MessagesRequestTranslator.md](core/llm-server/server/anthropic/MessagesRequestTranslator.md): Messages request to chat-completions body; has `translate`. - [core/llm-server/server/anthropic/MessageTurnTranslator.md](core/llm-server/server/anthropic/MessageTurnTranslator.md): one Messages turn to chat-completions messages; has `translate`. - [core/llm-server/server/anthropic/ToolDefinitionTranslator.md](core/llm-server/server/anthropic/ToolDefinitionTranslator.md): Anthropic tools to OpenAI functions; has `apply`, `functions`. - [core/llm-server/server/anthropic/AnthropicContent.md](core/llm-server/server/anthropic/AnthropicContent.md): system, tool_result and image content conversion; has `systemText`, `toolResultText`, `imagePart`. - [core/llm-server/server/anthropic/AnthropicIds.md](core/llm-server/server/anthropic/AnthropicIds.md): mints message and tool-use ids; has `messageId`, `toolUseId`. - [core/llm-server/server/anthropic/StopReason.md](core/llm-server/server/anthropic/StopReason.md): finish_reason to stop_reason; has `forFinish`, `forReply`. - [core/llm-server/server/anthropic/MessagesResponseTranslator.md](core/llm-server/server/anthropic/MessagesResponseTranslator.md): non-stream reply to a Messages response; has `translate`. - [core/llm-server/server/anthropic/MessagesStreamTranslator.md](core/llm-server/server/anthropic/MessagesStreamTranslator.md): stream chunks to Messages events; has `chunk`, `end`, `error`. - [core/llm-server/server/anthropic/MessagesTokenCount.md](core/llm-server/server/anthropic/MessagesTokenCount.md): count_tokens estimate; has `estimate`. - [core/llm-server/server/anthropic/AnthropicError.md](core/llm-server/server/anthropic/AnthropicError.md): Anthropic-shaped errors; has `send`, `body`. - [core/llm-server/server/relay/UpstreamChatRequest.md](core/llm-server/server/relay/UpstreamChatRequest.md): posts to llama-server chat completions for the translated routes; has `open`, `collect`. - [core/llm-server/server/anthropic/MessagesCompletion.md](core/llm-server/server/anthropic/MessagesCompletion.md): one non-streaming Messages request; has `run`. - [core/llm-server/server/anthropic/MessagesStream.md](core/llm-server/server/anthropic/MessagesStream.md): relays upstream SSE as Messages events; has `run`. - [core/llm-server/server/relay/TranslatedChatStream.md](core/llm-server/server/relay/TranslatedChatStream.md): base for a translated streaming request (relays upstream SSE through a dialect translator); has `run`, `parseLine`. - [core/llm-server/server/relay/TranslatedChatCompletion.md](core/llm-server/server/relay/TranslatedChatCompletion.md): base for a translated non-streaming request; has `run`. - [core/llm-server/server/OpenAiResponsesRouter.md](core/llm-server/server/OpenAiResponsesRouter.md): Express controller for the local OpenAI Responses API (what Codex speaks); has `create`. - [core/llm-server/server/responses/ResponsesRequestTranslator.md](core/llm-server/server/responses/ResponsesRequestTranslator.md): Responses request to chat-completions body; has `translate`. - [core/llm-server/server/responses/ResponseInputTranslator.md](core/llm-server/server/responses/ResponseInputTranslator.md): Responses input items to chat-completions messages; has `translate`. - [core/llm-server/server/responses/ResponsesToolTranslator.md](core/llm-server/server/responses/ResponsesToolTranslator.md): Responses function and custom tools to chat functions; has `apply`. - [core/llm-server/server/responses/ResponsesContent.md](core/llm-server/server/responses/ResponsesContent.md): content parts and tool outputs to chat text and parts; has `text`, `userContent`, `toolOutput`. - [core/llm-server/server/responses/ResponsesIds.md](core/llm-server/server/responses/ResponsesIds.md): mints response, item and call ids; has `responseId`, `itemId`, `callId`. - [core/llm-server/server/responses/ResponsesError.md](core/llm-server/server/responses/ResponsesError.md): OpenAI-shaped errors and failure codes; has `send`, `body`, `failure`, `codeFor`. - [core/llm-server/server/responses/OutputItems.md](core/llm-server/server/responses/OutputItems.md): message, reasoning and tool-call output items; has `message`, `reasoning`, `toolCall`, `customInput`. - [core/llm-server/server/responses/ResponseEnvelope.md](core/llm-server/server/responses/ResponseEnvelope.md): the response object per lifecycle state; has `inProgress`, `finished`, `failed`, `echo`, `usage`. - [core/llm-server/server/responses/ResponsesResponseTranslator.md](core/llm-server/server/responses/ResponsesResponseTranslator.md): non-stream reply to a Responses object; has `translate`. - [core/llm-server/server/responses/ResponsesStreamTranslator.md](core/llm-server/server/responses/ResponsesStreamTranslator.md): stream chunks to Responses events; has `chunk`, `end`, `error`. - [core/llm-server/server/responses/StreamedOutputItem.md](core/llm-server/server/responses/StreamedOutputItem.md): base for one output item while it streams; has `open`, `append`, `close`. - [core/llm-server/server/responses/StreamedTextItem.md](core/llm-server/server/responses/StreamedTextItem.md): base for a streamed item with one text part; has `append`. - [core/llm-server/server/responses/StreamedMessageItem.md](core/llm-server/server/responses/StreamedMessageItem.md): the streamed assistant message item. - [core/llm-server/server/responses/StreamedReasoningItem.md](core/llm-server/server/responses/StreamedReasoningItem.md): the streamed reasoning item. - [core/llm-server/server/responses/StreamedToolCallItem.md](core/llm-server/server/responses/StreamedToolCallItem.md): the streamed function or custom tool call item; has `append`. - [core/llm-server/server/responses/ResponsesCompletion.md](core/llm-server/server/responses/ResponsesCompletion.md): one non-streaming Responses request; has `run`. - [core/llm-server/server/responses/ResponsesStream.md](core/llm-server/server/responses/ResponsesStream.md): relays upstream SSE as Responses events; has `run`. - [core/llm-server/server/chat/OpenAICompatAdapter.md](core/llm-server/server/chat/OpenAICompatAdapter.md): ChatAdapter for OpenAI-compatible servers; has `healthCheck`, `chat`, `protocolId`. - [core/llm-server/server/chat/OpenAiChatStream.md](core/llm-server/server/chat/OpenAiChatStream.md): state of one streaming completion; has `handleLines`, `end`, `fail`, `abort`, `summary`. - [core/llm-server/server/chat/OpenAiChatRequest.md](core/llm-server/server/chat/OpenAiChatRequest.md): SSE chat request with one dead-socket retry; has `start`. - [core/llm-server/server/chat/ChatErrorBody.md](core/llm-server/server/chat/ChatErrorBody.md): HTTP error body to one capped line; has `read`. - [core/llm-server/server/chat/TimingsUsage.md](core/llm-server/server/chat/TimingsUsage.md): OpenAI usage from llama.cpp timings; has `fromTimings`. - [core/llm-server/validation/CodeValidator.md](core/llm-server/validation/CodeValidator.md): dispatches code to the right language validator; has `validate`, `formatForModel`, `isSupported`, `register`. - [core/llm-server/validation/ValidationResult.md](core/llm-server/validation/ValidationResult.md): validation result shape and model-facing block; has `fromDiagnostics`, `formatForModel`. - [core/llm-server/gambit/GambitRunner.md](core/llm-server/gambit/GambitRunner.md): runs the model compatibility suite and returns the report; has `runGambit`, `loadSuite`. - [core/llm-server/gambit/GambitCapabilities.md](core/llm-server/gambit/GambitCapabilities.md): host capabilities for the gambit with skip reasons; has `detect`, `probeWeb`. - [core/llm-server/gambit/GambitConversation.md](core/llm-server/gambit/GambitConversation.md): plays one gambit task as a conversation; has `play`. - [core/llm-server/gambit/GambitBridgeTurn.md](core/llm-server/gambit/GambitBridgeTurn.md): live gambit turns through the agent bridge; has `create`. - [core/llm-server/models/AddonModelSetup.md](core/llm-server/models/AddonModelSetup.md): one-click add-on model setup; has `execute`, `isInstalled`, `sha256File`. - [core/llm-server/runtimes/LlmRuntimeDetector.md](core/llm-server/runtimes/LlmRuntimeDetector.md): LLM runtime detector, with the backend device check on installed GPU builds; has `shared`, `detectRuntimes`, `parseVersionOutput`, `listBackendDevices`. - [core/llm-server/runtimes/LlmRuntimeInstaller.md](core/llm-server/runtimes/LlmRuntimeInstaller.md): LLM runtime installer; has `shared`, `installRuntime`, `uninstallRuntime`, `fetchLatestRelease`. - [core/llm-server/chat/ToolGroups.md](core/llm-server/chat/ToolGroups.md): lazy tool-group registry, the prose tool doc surface; has `staticGroups`, `extGroupsFor`, `extToolDoc`, `groupForTool`, `buildInactiveRegistry`. - [core/llm-server/chat/tool-groups/ScheduleBounds.md](core/llm-server/chat/tool-groups/ScheduleBounds.md): schedule bounds in minutes for tool docs; has `rangeText`. - [core/llm-server/chat/tool-groups/ArtifactManuals.md](core/llm-server/chat/tool-groups/ArtifactManuals.md): artifact, live artifact and widget-data tool manuals (data). - [core/llm-server/chat/tool-groups/MediaManuals.md](core/llm-server/chat/tool-groups/MediaManuals.md): image, video and music tool manuals (data). - [core/llm-server/chat/tool-groups/UtilityManuals.md](core/llm-server/chat/tool-groups/UtilityManuals.md): web, knowledge base, validate and webhook tool manuals (data). - [core/llm-server/chat/tool-groups/ActivationManual.md](core/llm-server/chat/tool-groups/ActivationManual.md): activate_tools manual and inactive-registry text (data). - [core/llm-server/chat/tool-groups/StaticGroupTable.md](core/llm-server/chat/tool-groups/StaticGroupTable.md): built-in tool groups in render order (data). - [core/llm-server/chat/tool-groups/MergedSourceTable.md](core/llm-server/chat/tool-groups/MergedSourceTable.md): tool sources merged into one group; has `forSource`. - [core/llm-server/chat/tool-groups/ExtensionGroupBuilder.md](core/llm-server/chat/tool-groups/ExtensionGroupBuilder.md): discovered tools to lazy groups; has `groupsFor`, `toolDoc`. - [core/llm-server/chat/tool-groups/FenceExampleStripper.md](core/llm-server/chat/tool-groups/FenceExampleStripper.md): strips fence examples for native tool routes; has `strip`. - [core/llm-server/chat/tool-groups/GeneratedPayload.md](core/llm-server/chat/tool-groups/GeneratedPayload.md): bounce exemption for calls carrying generated content; has `carries`. - [core/llm-server/chat/ToolSchemas.md](core/llm-server/chat/ToolSchemas.md): OpenAI tools array for native-tool turns, the JSON doc surface; has `buildHarmonyTools`, `browserToolSchemas`, `pseudoToolSchemas`, `extToolSchemas`, `stubToolSchema`. - [core/llm-server/chat/tool-schemas/FunctionSchema.md](core/llm-server/chat/tool-schemas/FunctionSchema.md): builds one OpenAI function entry; has `build`, `nameOf`. - [core/llm-server/chat/tool-schemas/PseudoToolSchemaTable.md](core/llm-server/chat/tool-schemas/PseudoToolSchemaTable.md): pseudo-tool schemas (data). - [core/llm-server/chat/tool-schemas/BrowserToolSchemas.md](core/llm-server/chat/tool-schemas/BrowserToolSchemas.md): browser tool schemas from BrowserTools; has `build`, `paramToSchema`. - [core/llm-server/chat/tool-schemas/LazySchemaStubber.md](core/llm-server/chat/tool-schemas/LazySchemaStubber.md): stubs tools of inactive groups on the native path; has `apply`, `stubSchema`. - [core/llm-server/chat/tool-schemas/FenceFallbackDoc.md](core/llm-server/chat/tool-schemas/FenceFallbackDoc.md): fence contract for natively excluded tools; has `build`. - [core/llm-server/chat/Compaction.md](core/llm-server/chat/Compaction.md): summarizes older chat history to fit the context window; has `shouldCompact`, `findCutIndex`, `compact`, `buildSummaryPrompt`. - [core/llm-server/chat/compaction/SummaryDirective.md](core/llm-server/chat/compaction/SummaryDirective.md): compaction directive and summary frame; has `wrap`. - [core/llm-server/chat/compaction/TurnBoundary.md](core/llm-server/chat/compaction/TurnBoundary.md): where a compaction cut may land; has `isTurnStart`, `isCutPoint`. - [core/llm-server/chat/compaction/MessageText.md](core/llm-server/chat/compaction/MessageText.md): flattens message content to text; has `of`. - [core/llm-server/chat/ToolOutputTruncator.md](core/llm-server/chat/ToolOutputTruncator.md): bounds tool output with head and tail selection and optional spill; has `forSlotBudget`, `truncate`. - [core/llm-server/chat/tool-output/TruncationNotice.md](core/llm-server/chat/tool-output/TruncationNotice.md): the model-facing truncation notice; has `build`. - [core/llm-server/chat/SafeFetch.md](core/llm-server/chat/SafeFetch.md): SSRF-hardened fetch for the agent's headless web tools; has `fetch`. - [core/llm-server/chat/safe-fetch/AddressGuard.md](core/llm-server/chat/safe-fetch/AddressGuard.md): which addresses SafeFetch may connect to, including IP-literal hosts; has `isBlockedAddress`, `literalRefusal`, `lookup`. - [core/llm-server/chat/safe-fetch/FetchOptions.md](core/llm-server/chat/safe-fetch/FetchOptions.md): SafeFetch options and Chrome-identity headers; has `from`, `followsRedirects`, `headers`. - [core/llm-server/chat/safe-fetch/CappedBodyReader.md](core/llm-server/chat/safe-fetch/CappedBodyReader.md): reads a response body up to a byte cap; has `read`. - [core/llm-server/chat/safe-fetch/ResponseCharset.md](core/llm-server/chat/safe-fetch/ResponseCharset.md): maps a Content-Type charset to a Buffer encoding; has `encodingFor`. - [core/llm-server/chat/ScheduledTaskMode.md](core/llm-server/chat/ScheduledTaskMode.md): the scheduled-task chat mode; has `register`, `unregister`, `descriptor`, `buildTurn`. - [core/llm-server/chat/scheduled-task/ScheduledTaskPrompt.md](core/llm-server/chat/scheduled-task/ScheduledTaskPrompt.md): the scheduled-task setup system prompt; has `build`. - [core/llm-server/chat/scheduled-task/ScheduledTaskTools.md](core/llm-server/chat/scheduled-task/ScheduledTaskTools.md): the scheduled-task create, update and run tools; has `build`, `create`, `update`, `runNow`. - [core/llm-server/chat/scheduled-task/ScheduledTaskToolSpecs.md](core/llm-server/chat/scheduled-task/ScheduledTaskToolSpecs.md): scheduled-task tool names, descriptions and schemas; has `create`, `update`, `run`. - [core/llm-server/chat/scheduled-task/TaskInterval.md](core/llm-server/chat/scheduled-task/TaskInterval.md): task intervals as readable frequencies; has `describe`, `toMinutes`, `fromMinutes`. - [core/llm-server/chat/scheduled-task/TestRunOutcome.md](core/llm-server/chat/scheduled-task/TestRunOutcome.md): shapes a test run's result for the setup model; has `from`. - [core/llm-server/chat/triggers/TriggerFileTools.md](core/llm-server/chat/triggers/TriggerFileTools.md): run-scoped file tools for file triggers; has `build`. - [core/llm-server/chat/triggers/file-tools/ReadTriggerFileTool.md](core/llm-server/chat/triggers/file-tools/ReadTriggerFileTool.md): read_trigger_file, pinned to the watch folder; has `definition`, `execute`. - [core/llm-server/chat/triggers/file-tools/WriteWatchDirFileTool.md](core/llm-server/chat/triggers/file-tools/WriteWatchDirFileTool.md): write_file_in_watch_dir, pinned to the watch folder; has `definition`, `execute`. - [core/llm-server/chat/TriggerStore.md](core/llm-server/chat/TriggerStore.md): triggers with their runs, delivery log and instruction history; has trigger CRUD, versions, drift, memory, delivery and run methods. - [core/llm-server/chat/trigger-store/TriggerLifecycle.md](core/llm-server/chat/trigger-store/TriggerLifecycle.md): the derived trigger lifecycle; has `configHash`, `isArmable`, `statusOf`. - [core/llm-server/chat/trigger-store/TriggerSourceConfig.md](core/llm-server/chat/trigger-store/TriggerSourceConfig.md): normalises a trigger source per kind; has `normalize`, `merge`. - [core/llm-server/chat/trigger-store/TriggerActionConfig.md](core/llm-server/chat/trigger-store/TriggerActionConfig.md): normalises a trigger action; has `normalize`, `merge`. - [core/llm-server/chat/trigger-store/TriggerPatch.md](core/llm-server/chat/trigger-store/TriggerPatch.md): works out an update's merged config and armed state; has `apply`. - [core/llm-server/chat/trigger-store/TriggerPolicy.md](core/llm-server/chat/trigger-store/TriggerPolicy.md): base class for per-trigger policies; has `normalizeInto`, `of`. - [core/llm-server/chat/trigger-store/TriggerFailurePolicy.md](core/llm-server/chat/trigger-store/TriggerFailurePolicy.md): auto-pause, notify and retry policy; has `normalizeInto`, `of`. - [core/llm-server/chat/trigger-store/TriggerApprovalPolicy.md](core/llm-server/chat/trigger-store/TriggerApprovalPolicy.md): approval mode and always-allow list; has `normalizeInto`, `of`, `withTool`. - [core/llm-server/chat/trigger-store/TriggerMemoryPolicy.md](core/llm-server/chat/trigger-store/TriggerMemoryPolicy.md): persistent trigger memory policy; has `normalize`, `of`, `capNotes`. - [core/llm-server/chat/trigger-store/FailureStreak.md](core/llm-server/chat/trigger-store/FailureStreak.md): applies the failure policy to one finished run; has `after`. - [core/llm-server/chat/trigger-store/DriftRecord.md](core/llm-server/chat/trigger-store/DriftRecord.md): the stored payload-drift record; has `next`. - [core/llm-server/chat/trigger-store/CappedJson.md](core/llm-server/chat/trigger-store/CappedJson.md): JSON under a hard character cap; has `stringify`. - [core/llm-server/chat/trigger-store/DeliveryEventSummary.md](core/llm-server/chat/trigger-store/DeliveryEventSummary.md): compact event facts for the delivery log; has `compact`. - [core/llm-server/chat/trigger-store/HookToken.md](core/llm-server/chat/trigger-store/HookToken.md): webhook hook tokens; has `create`, `isWellFormed`. - [core/llm-server/chat/trigger-store/TriggerRowMapper.md](core/llm-server/chat/trigger-store/TriggerRowMapper.md): hydrates trigger table rows; has `trigger`, `run`, `delivery`, `version`. - [core/llm-server/chat/trigger-store/TriggerRunLog.md](core/llm-server/chat/trigger-store/TriggerRunLog.md): a trigger's fire history; has `start`, `finish`, `get`, `list`, `hasDedupeKey`, `pruneTranscripts`. - [core/llm-server/chat/trigger-store/TriggerDeliveryLog.md](core/llm-server/chat/trigger-store/TriggerDeliveryLog.md): per-trigger inbound delivery log; has `record`, `update`, `get`, `list`, `counts`. - [core/llm-server/chat/trigger-store/TriggerVersionHistory.md](core/llm-server/chat/trigger-store/TriggerVersionHistory.md): trigger instruction version history; has `record`, `list`, `info`, `find`, `markTested`. - [core/llm-server/chat/trigger-store/TriggerRepository.md](core/llm-server/chat/trigger-store/TriggerRepository.md): plain SQL over the triggers table; has `insert`, `get`, `getByToken`, `list`, `delete`, `updateConfig`. - [core/llm-server/chat/trigger-store/TriggerRunRepository.md](core/llm-server/chat/trigger-store/TriggerRunRepository.md): plain SQL over trigger runs; has `insert`, `finish`, `get`, `list`, `hasDedupeKey`. - [core/llm-server/chat/trigger-store/TriggerDeliveryRepository.md](core/llm-server/chat/trigger-store/TriggerDeliveryRepository.md): plain SQL over trigger deliveries; has `insert`, `update`, `get`, `list`, `counts`, `prune`. - [core/llm-server/chat/trigger-store/TriggerVersionRepository.md](core/llm-server/chat/trigger-store/TriggerVersionRepository.md): plain SQL over trigger versions; has `insert`, `list`, `latest`, `get`, `markTested`, `prune`. - [core/llm-server/server/LaunchPlanner.md](core/llm-server/server/LaunchPlanner.md): plans a llama-server launch (VRAM fit, KV and expert placement, splits, speculation, serving flags); has `plan`. - [core/llm-server/server/launch/KvCacheType.md](core/llm-server/server/launch/KvCacheType.md): accepted KV cache types and bytes per element; has `normalize`, `elementBytes`, `isQuantized`. - [core/llm-server/server/launch/SlidingWindowLayout.md](core/llm-server/server/launch/SlidingWindowLayout.md): which layers cache only a sliding window; has `of`. - [core/llm-server/server/launch/HybridAttentionLayout.md](core/llm-server/server/launch/HybridAttentionLayout.md): which hybrid-attention blocks keep KV; has `of`. - [core/llm-server/server/launch/KvCacheSizer.md](core/llm-server/server/launch/KvCacheSizer.md): sliding-window and hybrid-aware KV cache bytes; has `perLayer`, `total`, `atLayer`. - [core/llm-server/server/launch/GpuOverhead.md](core/llm-server/server/launch/GpuOverhead.md): per-card reserve and fixed GPU scratch; has `fixed`. - [core/llm-server/server/launch/DraftBranchCost.md](core/llm-server/server/launch/DraftBranchCost.md): VRAM of an MTP or drafter branch; has `bytes`. - [core/llm-server/server/launch/FullOffloadCost.md](core/llm-server/server/launch/FullOffloadCost.md): full-offload VRAM price; has `bytes`, `noHeaderBytes`. - [core/llm-server/server/launch/PartialOffloadSizer.md](core/llm-server/server/launch/PartialOffloadSizer.md): sizes a partial GPU layer split; has `size`. - [core/llm-server/server/launch/GpuInventory.md](core/llm-server/server/launch/GpuInventory.md): local and remote GPUs to budget against; has `fromDiagnostics`, `remoteDevices`, `largestIndex`. - [core/llm-server/server/launch/TensorParallelRatio.md](core/llm-server/server/launch/TensorParallelRatio.md): bandwidth-weighted tensor-parallel ratio; has `compute`. - [core/llm-server/server/launch/MoeLayerSplit.md](core/llm-server/server/launch/MoeLayerSplit.md): byte-balanced multi-GPU split for MoE expert offload; has `balance`. - [core/llm-server/server/launch/MoeGpuCachePlan.md](core/llm-server/server/launch/MoeGpuCachePlan.md): every expert in RAM plus a `--moe-cache-mib` device cache instead of the layer fill; has `resolve`. - [core/llm-server/server/launch/LayerFillSplit.md](core/llm-server/server/launch/LayerFillSplit.md): fastest-card-first layer split; has `compute`, `describe`. - [core/llm-server/server/launch/RpcSplitRatio.md](core/llm-server/server/launch/RpcSplitRatio.md): split across remote RPC GPUs; has `compute`. - [core/llm-server/server/launch/PromptCacheRam.md](core/llm-server/server/launch/PromptCacheRam.md): sizes the prompt cache in RAM; has `resolve`. - [core/llm-server/server/launch/ModelFiles.md](core/llm-server/server/launch/ModelFiles.md): a model's weights, projector, drafter and header for planning. - [core/llm-server/server/launch/RuntimeFlags.md](core/llm-server/server/launch/RuntimeFlags.md): which flags a runtime build accepts; has `accepts`. - [core/llm-server/server/launch/KvSettings.md](core/llm-server/server/launch/KvSettings.md): context clamp and KV precision for a launch; has `forceF16`. - [core/llm-server/server/launch/SlotPlan.md](core/llm-server/server/launch/SlotPlan.md): parallel slots and cache reuse; has `resolve`. - [core/llm-server/server/launch/SpeculationPlan.md](core/llm-server/server/launch/SpeculationPlan.md): speculative decoding choice and budget; has `resolve`. - [core/llm-server/server/launch/VisionProjectorPlan.md](core/llm-server/server/launch/VisionProjectorPlan.md): whether the vision projector loads; has `resolve`. - [core/llm-server/server/launch/VramBudget.md](core/llm-server/server/launch/VramBudget.md): usable VRAM including remote peers and the cap; has `resolve`. - [core/llm-server/server/launch/TensorParallelGate.md](core/llm-server/server/launch/TensorParallelGate.md): whether tensor split is eligible; has `resolve`. - [core/llm-server/server/launch/LaunchCostInput.md](core/llm-server/server/launch/LaunchCostInput.md): full-offload cost inputs for a launch; has `build`, `noHeader`. - [core/llm-server/server/launch/MoeExpertPlan.md](core/llm-server/server/launch/MoeExpertPlan.md): MoE expert CPU offload and spare-VRAM fill; has `resolve`. - [core/llm-server/server/launch/OffloadDecision.md](core/llm-server/server/launch/OffloadDecision.md): full, partial or CPU decision; has `resolve`. - [core/llm-server/server/launch/SplitPlan.md](core/llm-server/server/launch/SplitPlan.md): which tensor-split owner applies; has `resolve`. - [core/llm-server/server/launch/LoadingPolicy.md](core/llm-server/server/launch/LoadingPolicy.md): mmap versus bulk read and prompt cache; has `resolve`. - [core/llm-server/server/launch/FamilyTuning.md](core/llm-server/server/launch/FamilyTuning.md): family sampler, template and drafter args; has `resolve`. - [core/llm-server/server/launch/LaunchArgsBuilder.md](core/llm-server/server/launch/LaunchArgsBuilder.md): assembles the llama-server argv in order; has `build`. - [core/llm-server/server/launch/PlacementNotes.md](core/llm-server/server/launch/PlacementNotes.md): plan notes on runtime, VRAM need and offload; has `build`. - [core/llm-server/server/launch/ModelFeatureNotes.md](core/llm-server/server/launch/ModelFeatureNotes.md): plan notes on KV, attention and speculation; has `build`. - [core/llm-server/server/launch/ServingNotes.md](core/llm-server/server/launch/ServingNotes.md): plan notes on slots, caches and splits; has `build`. - [core/llm-server/server/launch/PlanSummary.md](core/llm-server/server/launch/PlanSummary.md): the plan object returned with the argv; has `build`. - [core/llm-server/models/FitVramMath.md](core/llm-server/models/FitVramMath.md): shared pre-download KV and overhead arithmetic; has `kvBytes`, `kvBytesAt8k`. - [core/llm-server/models/FitClassifier.md](core/llm-server/models/FitClassifier.md): pre-download fit badge with predicted speed; has `classify`. - [core/llm-server/models/MlxFitClassifier.md](core/llm-server/models/MlxFitClassifier.md): fit badge for MLX on unified memory; has `classify`. - [core/llm-server/models/Recommender.md](core/llm-server/models/Recommender.md): onboarding model, quant, context and runtime pick; has `recommend`. - [core/llm-server/models/RecommenderRuntimePicker.md](core/llm-server/models/RecommenderRuntimePicker.md): runtime choice for a machine; has `pick`. - [core/llm-server/models/RecommenderRationale.md](core/llm-server/models/RecommenderRationale.md): wizard explanation text for a recommendation; has `build`. - [core/llm-server/LlmModelsScanner.md](core/llm-server/LlmModelsScanner.md): scans the LLM models folder into classified entries with parsed headers; has `shared`, `scan`. - [core/llm-server/scanner/GgufParseWorkerClient.md](core/llm-server/scanner/GgufParseWorkerClient.md): parses GGUF headers on a lazy, idle-terminated worker; has `parse`, `stop`. - [core/llm-server/scanner/GgufHeaderCache.md](core/llm-server/scanner/GgufHeaderCache.md): persisted GGUF header memo; has `get`, `flush`. - [core/llm-server/scanner/ShardScanMerger.md](core/llm-server/scanner/ShardScanMerger.md): merges split-model shard scans into one layout; has `merge`. - [core/llm-server/scanner/ModelDirWalker.md](core/llm-server/scanner/ModelDirWalker.md): capped folder walk collecting model candidates; has `walk`. - [core/llm-server/scanner/GgufModelGrouper.md](core/llm-server/scanner/GgufModelGrouper.md): groups GGUF files into models with shards, projectors and drafters, roles from the resolver; has `group`, `roleOf`. - [core/llm-server/scanner/GgufRoleResolver.md](core/llm-server/scanner/GgufRoleResolver.md): per-file role before grouping, header first and file name as fallback; has `resolve`, `resolveOne`, `fromFilename`. - [core/llm-server/scanner/GgufCompanionPairer.md](core/llm-server/scanner/GgufCompanionPairer.md): pairs a projector, drafter or MTP head with one model family or none, with the draft size ratio; has `pair`, `families`, `isDraftOf`, `keysOf`. - [core/llm-server/scanner/MlxModelBuilder.md](core/llm-server/scanner/MlxModelBuilder.md): builds MLX model entries; has `build`, `readConfigFacts`. - [core/llm-server/scanner/AddonModelBuilder.md](core/llm-server/scanner/AddonModelBuilder.md): builds add-on model entries; has `build`. - [core/llm-server/scanner/ModelSidecars.md](core/llm-server/scanner/ModelSidecars.md): finds metadata files beside model weights; has `attach`, `forModel`. - [core/llm-server/scanner/ModelClassifier.md](core/llm-server/scanner/ModelClassifier.md): model formats, runtimes and MTP capability; has `classify`, `augmentFromGguf`. - [core/llm-server/scanner/GgufHeaderAttacher.md](core/llm-server/scanner/GgufHeaderAttacher.md): reads a model's GGUF header into its entry; has `attach`, `factsOf`. - [core/llm-server/scanner/BoundedParallel.md](core/llm-server/scanner/BoundedParallel.md): ordered async map with a concurrency cap; has `map`. - [core/llm-server/models/AutoPlanner.md](core/llm-server/models/AutoPlanner.md): Automatic Local Setup planner for chat, image and music; has `plan`. - [core/llm-server/models/auto-plan/AutoPlanRanking.md](core/llm-server/models/auto-plan/AutoPlanRanking.md): shared ladders and pick ranking for setup; has `modelScore`, `vramNeedOf`, `predictedTps`, `tierOf`. - [core/llm-server/models/auto-plan/LlmTierPicker.md](core/llm-server/models/auto-plan/LlmTierPicker.md): four-tier chat model pick; has `pick`. - [core/llm-server/models/auto-plan/AutoPlanMachine.md](core/llm-server/models/auto-plan/AutoPlanMachine.md): hardware budget to planning facts; has `describe`. - [core/llm-server/models/auto-plan/CoexistencePicker.md](core/llm-server/models/auto-plan/CoexistencePicker.md): per-card chat and image coexistence picks; has `coexists`, `residentPick`, `moePick`. - [core/llm-server/models/auto-plan/ImagePlacementPlanner.md](core/llm-server/models/auto-plan/ImagePlacementPlanner.md): image placement decision with the chat trade-off; has `decide`. - [core/llm-server/models/auto-plan/MusicLegPlanner.md](core/llm-server/models/auto-plan/MusicLegPlanner.md): music eligibility and pool joining; has `plan`. - [core/llm-server/models/auto-plan/LlmRecommendation.md](core/llm-server/models/auto-plan/LlmRecommendation.md): chat pick to install recommendation; has `build`. - [core/llm-server/models/auto-plan/RamPinAdvice.md](core/llm-server/models/auto-plan/RamPinAdvice.md): RAM pin recommendation for swap pools; has `advise`. - [core/llm-server/models/auto-plan/PlanSummary.md](core/llm-server/models/auto-plan/PlanSummary.md): setup plan preview lines; has `build`. - [core/llm-server/models/CatalogBuilder.md](core/llm-server/models/CatalogBuilder.md): builds the curated catalog from HF repo facts; has `build`, `buildEntry`. - [core/llm-server/models/LiveCatalog.md](core/llm-server/models/LiveCatalog.md): cached live catalog with a static fallback; has `shared`, `get`, `reset`. - [core/llm-server/agent/AgentRunner.md](core/llm-server/agent/AgentRunner.md): headless agent tool loop facade (moved from the ai-chat extension into core); has `run`, `buildSystemPrompt`. - [core/llm-server/agent/AgentRun.md](core/llm-server/agent/AgentRun.md): one execution of the agent loop; has `execute`. - [core/llm-server/agent/AgentHistory.md](core/llm-server/agent/AgentHistory.md): run messages plus tool-history and carried-note ledgers; has `push`, `pushToolResult`, `evictToolHistory`, `applyCompaction`. - [core/llm-server/agent/AgentLoopText.md](core/llm-server/agent/AgentLoopText.md): model-facing agent loop strings (data). - [core/llm-server/agent/AgentPromptText.md](core/llm-server/agent/AgentPromptText.md): browser tool lines for the agent prompt (data); has `browserToolLines`. - [core/llm-server/agent/AgentSystemPromptBuilder.md](core/llm-server/agent/AgentSystemPromptBuilder.md): builds the agent system prompt with overrides; has `build`. - [core/llm-server/agent/InterstitialWait.md](core/llm-server/agent/InterstitialWait.md): waits out a "please wait" / redirecting / browser-check page after a navigation instead of letting the agent navigate again; has `settle`, `isInterstitial`. - [core/llm-server/agent/AgentToolExecutor.md](core/llm-server/agent/AgentToolExecutor.md): executes one tool call on the working tab; has `execute`. - [core/llm-server/agent/CarriedNotesLedger.md](core/llm-server/agent/CarriedNotesLedger.md): eviction order for carried reasoning notes; has `track`, `remap`. - [core/llm-server/agent/CarriedReasoning.md](core/llm-server/agent/CarriedReasoning.md): carried-reasoning budgets; has `resolveBudget`, `tail`. - [core/llm-server/agent/CompletionSender.md](core/llm-server/agent/CompletionSender.md): agent completions with warm-up retry; has `send`, `sendFirst`, `summarize`. - [core/llm-server/agent/CompletionVerifier.md](core/llm-server/agent/CompletionVerifier.md): tracks page actions that had no effect; has `next`, `observe`, `takeNudge`. - [core/llm-server/agent/ContextOverflow.md](core/llm-server/agent/ContextOverflow.md): recognises context-overflow errors; has `isExceeded`, `parseCounts`. - [core/llm-server/agent/FinalAnswerGate.md](core/llm-server/agent/FinalAnswerGate.md): retry, stall and verify nudges before a final answer; has `review`. - [core/llm-server/agent/LargeStringShrinker.md](core/llm-server/agent/LargeStringShrinker.md): cuts the largest strings of an over-budget result; has `shrink`. - [core/llm-server/agent/MidTurnCompactor.md](core/llm-server/agent/MidTurnCompactor.md): proactive and reactive in-turn compaction; has `compactIfOverTrigger`, `compact`. - [core/llm-server/agent/OverflowRecovery.md](core/llm-server/agent/OverflowRecovery.md): evict, shrink, compact and retry after an overflow; has `recover`. - [core/llm-server/agent/RunClock.md](core/llm-server/agent/RunClock.md): agent run wall clock; has `expired`, `remainingMs`, `extend`. - [core/llm-server/agent/RunOptions.md](core/llm-server/agent/RunOptions.md): agent run options with defaults; has `normalize`. - [core/llm-server/agent/RunSummary.md](core/llm-server/agent/RunSummary.md): one-sentence agent run summary; has `generate`. - [core/llm-server/agent/SystemPromptRenderer.md](core/llm-server/agent/SystemPromptRenderer.md): renders the base agent system prompt; has `render`. - [core/llm-server/agent/SystemPromptText.md](core/llm-server/agent/SystemPromptText.md): base agent system prompt wording (data). - [core/llm-server/agent/TabLookup.md](core/llm-server/agent/TabLookup.md): tab description, URL and element digest for the agent; has `describe`, `currentUrl`, `attachDigest`. - [core/llm-server/agent/ToolBatchParser.md](core/llm-server/agent/ToolBatchParser.md): all tool calls of a reply in model order; has `parse`. - [core/llm-server/agent/ToolBatchRunner.md](core/llm-server/agent/ToolBatchRunner.md): runs one reply's tool calls in parallel or exclusively; has `run`. - [core/llm-server/agent/ToolCallGuard.md](core/llm-server/agent/ToolCallGuard.md): allow-list rejection before a tool call; has `check`. - [core/llm-server/agent/ToolHistoryBudget.md](core/llm-server/agent/ToolHistoryBudget.md): tool-history character budgets; has `forSlot`. - [core/llm-server/agent/ToolHistoryEvictor.md](core/llm-server/agent/ToolHistoryEvictor.md): size-weighted tool-result eviction; has `evict`. - [core/llm-server/agent/ToolResultCompactor.md](core/llm-server/agent/ToolResultCompactor.md): tool result to model text; has `compact`. - [core/llm-server/agent/UnfulfilledIntent.md](core/llm-server/agent/UnfulfilledIntent.md): detects announce-only replies; has `matches`. - [core/llm-server/agent/WorkTab.md](core/llm-server/agent/WorkTab.md): the agent's working tab; has `open`, `ensure`, `closeIfOwned`. - [core/llm-server/server/LlmRuntimeServer.md](core/llm-server/server/LlmRuntimeServer.md): supervises the chat model's llama-server; has `findFreePort`, `baseUrl`, `markDirty`, `ensureStopped` plus the base API. - [core/llm-server/server/LlmCapsProbe.md](core/llm-server/server/LlmCapsProbe.md): learns how the loaded model's template controls thinking; has `execute`, `emptyCaps`. - [core/llm-server/router/RouterRuntimeServer.md](core/llm-server/router/RouterRuntimeServer.md): supervises the tool-group router's CPU llama-server; has `findFreePort` plus the base API. - [core/llm-server/server/chat/ChatAdapterRegistry.md](core/llm-server/server/chat/ChatAdapterRegistry.md): maps a runtime protocol to its chat adapter; has `createAdapterFor`, `listProtocols`. - [core/llm-server/server/PlanFor.md](core/llm-server/server/PlanFor.md): picks and runs the right launch planner; has `plan`, `isCustomLaunch`, `isMlx`. - [core/llm-server/server/LocalApiServer.md](core/llm-server/server/LocalApiServer.md): localhost OpenAI- (chat and Responses) and Anthropic-compatible endpoint; has `setEnabled`, `setPort`, `start`, `stop`, `getConfig`, `getBaseUrl`. - [core/llm-server/server/LocalApiUpstream.md](core/llm-server/server/LocalApiUpstream.md): what the local API sees of the running model; has `current`, `getThinking`, `listModels`. - [core/llm-server/server/LocalApiListener.md](core/llm-server/server/LocalApiListener.md): loopback-only HTTP listener; has `start`, `stop`, `isRunning`, `address`. - [core/llm-server/chat/ToolRequiredArgs.md](core/llm-server/chat/ToolRequiredArgs.md): enforces required tool arguments at dispatch; has `missing`, `softMissing`, `refusal`, `withSoftMissingNote`. - [core/llm-server/chat/triggers/WebhookSource.md](core/llm-server/chat/triggers/WebhookSource.md): the inbound webhook receiver for triggers; has `router`. - [core/llm-server/chat/triggers/webhook/WebhookDelivery.md](core/llm-server/chat/triggers/webhook/WebhookDelivery.md): one webhook delivery from rate limit and signature check to fire; has `handle`. - [core/llm-server/chat/triggers/webhook/WebhookFireReply.md](core/llm-server/chat/triggers/webhook/WebhookFireReply.md): trigger fire outcome to HTTP reply; has `refused`, `ack`, `result`. - [core/llm-server/chat/triggers/webhook/WebhookEvent.md](core/llm-server/chat/triggers/webhook/WebhookEvent.md): event record from a webhook request with header redaction; has `build`, `safeHeaders`. - [core/llm-server/chat/triggers/webhook/WebhookBodyParser.md](core/llm-server/chat/triggers/webhook/WebhookBodyParser.md): parses raw webhook bodies by content type; has `parse`. - [core/llm-server/chat/triggers/webhook/WindowCounter.md](core/llm-server/chat/triggers/webhook/WindowCounter.md): sliding-window counters for rate limits and lockouts; has `hit`, `count`. - [core/llm-server/chat/WebLookup.md](core/llm-server/chat/WebLookup.md): the headless web_search tool (search, paged page reads); has `run`. - [core/llm-server/chat/web-tools/LookupRequest.md](core/llm-server/chat/web-tools/LookupRequest.md): validated web_search arguments; has `parse`. - [core/llm-server/chat/web-tools/LookupReply.md](core/llm-server/chat/web-tools/LookupReply.md): result shapes and actionable error voice; has `ok`, `fail`, `reasonOf`, `today`. - [core/llm-server/chat/web-tools/WebSession.md](core/llm-server/chat/web-tools/WebSession.md): one run's web state (results, read pages). - [core/llm-server/chat/web-tools/ResultMemory.md](core/llm-server/chat/web-tools/ResultMemory.md): numbered search results and near-match lookup; has `remember`, `byNumber`, `nearest`. - [core/llm-server/chat/web-tools/PageCache.md](core/llm-server/chat/web-tools/PageCache.md): short-lived per-run cache of read pages; has `get`, `put`. - [core/llm-server/chat/web-tools/ReadBudget.md](core/llm-server/chat/web-tools/ReadBudget.md): characters per page part from the slot size; has `partChars`. - [core/llm-server/chat/web-tools/PageReader.md](core/llm-server/chat/web-tools/PageReader.md): reads one URL headless, then in a silent tab; has `read`. - [core/llm-server/chat/web-tools/PageReport.md](core/llm-server/chat/web-tools/PageReport.md): one bounded page reply with parts, find excerpts and continuation; has `render`. - [core/llm-server/chat/web-tools/PageNotes.md](core/llm-server/chat/web-tools/PageNotes.md): redirect, cut-download and thin-page caveats; has `collect`. - [core/llm-server/chat/web-tools/TextPager.md](core/llm-server/chat/web-tools/TextPager.md): splits text into numbered parts; has `count`, `part`, `partAt`. - [core/llm-server/chat/web-tools/MatchExcerpts.md](core/llm-server/chat/web-tools/MatchExcerpts.md): excerpts around term matches; has `extract`. - [core/llm-server/chat/web-tools/BoilerplateTrimmer.md](core/llm-server/chat/web-tools/BoilerplateTrimmer.md): trims link-dense menus and footers from page edges; has `trim`, `isFurniture`. - [core/llm-server/chat/web-tools/ResponseText.md](core/llm-server/chat/web-tools/ResponseText.md): response body to model text; has `isHtml`, `toText`. - [core/llm-server/chat/web-tools/PageClassifier.md](core/llm-server/chat/web-tools/PageClassifier.md): bot walls, nothing-found, prose, tab-worthy failures; has `isChallenge`, `isNothingFound`, `isProse`, `tabMightSucceed`. - [core/llm-server/chat/web-tools/SilentTab.md](core/llm-server/chat/web-tools/SilentTab.md): renders a page in a background tab that never takes focus; has `make`, `render`. - [core/llm-server/chat/web-tools/SearchRunner.md](core/llm-server/chat/web-tools/SearchRunner.md): web search over a ladder of engines, topped up without repeats; has `run`. - [core/llm-server/chat/web-tools/ResultList.md](core/llm-server/chat/web-tools/ResultList.md): de-duplicated numbered result list; has `add`, `reply`. - [core/llm-server/chat/web-tools/UrlIdentity.md](core/llm-server/chat/web-tools/UrlIdentity.md): when two URLs are the same page; has `key`, `same`. - [core/llm-server/chat/web-tools/EditDistance.md](core/llm-server/chat/web-tools/EditDistance.md): edit distance for URLs; has `levenshtein`, `similarity`. - [core/llm-server/chat/web-tools/SearchEngine.md](core/llm-server/chat/web-tools/SearchEngine.md): base class for scraped search engines; has `searchUrl`, `parse`. - [core/llm-server/chat/web-tools/DuckDuckGoEngine.md](core/llm-server/chat/web-tools/DuckDuckGoEngine.md): DuckDuckGo HTML search engine; has `searchUrl`, `parse`. - [core/llm-server/chat/web-tools/BingEngine.md](core/llm-server/chat/web-tools/BingEngine.md): Bing fallback search engine; has `searchUrl`, `parse`. - [core/llm-server/chat/TriggerMode.md](core/llm-server/chat/TriggerMode.md): the trigger setup chat mode; has `register`, `unregister`, `descriptor`, `buildTurn`. - [core/llm-server/chat/trigger-mode/TriggerModePrompt.md](core/llm-server/chat/trigger-mode/TriggerModePrompt.md): trigger setup system prompt; has `build`. - [core/llm-server/chat/trigger-mode/TriggerModeGuide.md](core/llm-server/chat/trigger-mode/TriggerModeGuide.md): fixed trigger setup prompt text; has `lines`. - [core/llm-server/chat/trigger-mode/TriggerStateView.md](core/llm-server/chat/trigger-mode/TriggerStateView.md): a trigger's state for the model; has `build`. - [core/llm-server/chat/trigger-mode/TriggerFacts.md](core/llm-server/chat/trigger-mode/TriggerFacts.md): shared model-facing trigger facts; has `describeStatus`, `hookUrls`, `oneLine`. - [core/llm-server/chat/trigger-mode/TriggerTestOutcome.md](core/llm-server/chat/trigger-mode/TriggerTestOutcome.md): trigger test result for the model; has `from`. - [core/llm-server/chat/trigger-mode/TriggerModeServices.md](core/llm-server/chat/trigger-mode/TriggerModeServices.md): late-bound collaborators for trigger mode; has `runner`, `pageSource`, `hookBaseUrls`. - [core/llm-server/chat/trigger-mode/TriggerPromptExtras.md](core/llm-server/chat/trigger-mode/TriggerPromptExtras.md): live context for the trigger setup prompt; has `build`. - [core/llm-server/chat/trigger-mode/TriggerToolSpecs.md](core/llm-server/chat/trigger-mode/TriggerToolSpecs.md): trigger setup tool schemas; has `create`, `update`, `test`. - [core/llm-server/chat/trigger-mode/TriggerGatingParams.md](core/llm-server/chat/trigger-mode/TriggerGatingParams.md): gating parameters to source keys; has `toSource`. - [core/llm-server/chat/trigger-mode/TriggerCreateSource.md](core/llm-server/chat/trigger-mode/TriggerCreateSource.md): a new trigger's source per kind; has `kindOf`, `build`. - [core/llm-server/chat/trigger-mode/TriggerUpdatePatch.md](core/llm-server/chat/trigger-mode/TriggerUpdatePatch.md): update_trigger store patch; has `build`, `hasChanges`. - [core/llm-server/chat/trigger-mode/TriggerTools.md](core/llm-server/chat/trigger-mode/TriggerTools.md): the trigger setup tool handlers; has `build`, `create`, `test`, `update`. - [core/llm-server/chat/trigger-mode/TriggerSummary.md](core/llm-server/chat/trigger-mode/TriggerSummary.md): compact trigger summary; has `of`. - [core/llm-server/chat/ApprovalGate.md](core/llm-server/chat/ApprovalGate.md): which agent tool calls need a human yes, the refusal text, and run-grant families; has `requiresApproval`, `mutatingNamesOf`, `deniedResult`, `deniedCommandResult`, `familyOf`, `grantKeyOf`. - [core/llm-server/chat/approval/ApprovalPolicy.md](core/llm-server/chat/approval/ApprovalPolicy.md): whether the approval gate is on for a run; has `resolve`, `resolveRun`. - [core/llm-server/chat/approval/CommandCallAssessor.md](core/llm-server/chat/approval/CommandCallAssessor.md): shell-command verdict from the classifier and write boundary; has `assess`, `isEnabled`, `isCommandTool`. - [core/llm-server/chat/approval/ApprovalAttention.md](core/llm-server/chat/approval/ApprovalAttention.md): flashes the window and notifies when an approval card waits while LumaBrowser is not focused; has `request`, `clear`. - [core/llm-server/chat/approval/CallDescriber.md](core/llm-server/chat/approval/CallDescriber.md): the approval card's description of a call; has `describe`. - [core/llm-server/chat/approval/DesktopCallText.md](core/llm-server/chat/approval/DesktopCallText.md): approval card sentences for the desktop_* tools; has `launch`, `click`, `type`, `pressKey`, `scroll`, `DESCRIBERS`. - [core/llm-server/chat/approval/ConversationGrants.md](core/llm-server/chat/approval/ConversationGrants.md): carries the desktop control grant across a conversation's turns until 30 idle minutes; has `live`, `touch`, `runApprovals`. - [core/llm-server/chat/approval/CardText.md](core/llm-server/chat/approval/CardText.md): one-line clipping for approval cards; has `clip`. - [core/llm-server/chat/LiveApi.md](core/llm-server/chat/LiveApi.md): host side of a live artifact's `luma` bridge (page fetch, open tab, guarded extension API calls); has `fetchPage`, `openTab`, `extCall`. - [core/llm-server/chat/live-api/FetchQueue.md](core/llm-server/chat/live-api/FetchQueue.md): concurrency and queue cap for live page fetches; has `run`. - [core/llm-server/chat/live-api/LivePageRequest.md](core/llm-server/chat/live-api/LivePageRequest.md): validates live page fetch parameters; has `from`, `validHttpUrl`. - [core/llm-server/chat/live-api/LivePageFetch.md](core/llm-server/chat/live-api/LivePageFetch.md): one live page read, headless then silent tab, never for refused private addresses; has `execute`. - [core/llm-server/ContextEstimator.md](core/llm-server/ContextEstimator.md): per-model context ladders for the chat picker and the Setup fit matrix; has `buildLocalModelOptions`, `measuredComboVram`, `ctxLabel`. - [core/llm-server/context/ContextLadder.md](core/llm-server/context/ContextLadder.md): shared context rungs and labels; has `label`, `rungsFor`. - [core/llm-server/context/FitRows.md](core/llm-server/context/FitRows.md): lookups over saved fit-test rows; has `okRow`, `anyRow`, `measuredComboVram`. - [core/llm-server/context/RungEstimator.md](core/llm-server/context/RungEstimator.md): planner estimate for one context size; has `estimate`, `predictedTps`. - [core/llm-server/context/RungTip.md](core/llm-server/context/RungTip.md): hover text for one context rung; has `for`. - [core/llm-server/context/ContextRungBuilder.md](core/llm-server/context/ContextRungBuilder.md): one context rung from a measurement or an estimate; has `build`. - [core/llm-server/context/AutoContextPick.md](core/llm-server/context/AutoContextPick.md): auto-picked context ladder and default; has `ladder`, `recommendedTokens`. - [core/llm-server/context/ModelContextEntry.md](core/llm-server/context/ModelContextEntry.md): one model's chat picker entry; has `build`, `isChattable`. - [core/llm-server/context/EffectiveContext.md](core/llm-server/context/EffectiveContext.md): the per-slot context window in effect; has `resolve`, `clampSlots`. - [core/llm-server/server/VramWatchdog.md](core/llm-server/server/VramWatchdog.md): free-VRAM pressure watchdog over the nvidia-smi stream, degraded rather than disabled on bad readings, with optional idle unload; has `start`, `stop`, `tick`, `dismiss`, `getState`. - [core/llm-server/server/vram-watchdog/VramBand.md](core/llm-server/server/vram-watchdog/VramBand.md): normal, low or critical VRAM against the reserve; has `of`. - [core/llm-server/server/vram-watchdog/LedgerCards.md](core/llm-server/server/vram-watchdog/LedgerCards.md): watched and loading cards from the VRAM ledger; has `fromLedger`. - [core/llm-server/server/vram-watchdog/WatchedCard.md](core/llm-server/server/vram-watchdog/WatchedCard.md): one card's VRAM pressure state; has `sample`, `commit`, `dismiss`, `unloadDue`. - [core/llm-server/server/vram-watchdog/VramPressureText.md](core/llm-server/server/vram-watchdog/VramPressureText.md): VRAM pressure banner text; has `describe`. - [core/llm-server/rampin/LlmPinTarget.md](core/llm-server/rampin/LlmPinTarget.md): default chat model to the RAM-pin file list; has `resolve`. - [core/llm-server/router/GroupRouterService.md](core/llm-server/router/GroupRouterService.md): opt-in resident CPU tool-group router; has `apply`, `classify`, `downloadModel`, `getStatus`, `stop`. - [core/llm-server/router/GroupRouterSettings.md](core/llm-server/router/GroupRouterSettings.md): tool-group router settings and model location; has `isEnabled`, `setEnabled`, `modelPath`. - [core/llm-server/router/RouterModelDownloader.md](core/llm-server/router/RouterModelDownloader.md): verified router model download; has `download`, `cancel`. - [core/llm-server/router/RouterLaunchPlan.md](core/llm-server/router/RouterLaunchPlan.md): CPU-only router launch plan; has `pickRuntime`, `build`. - [core/llm-server/router/RouterClient.md](core/llm-server/router/RouterClient.md): one grammar-constrained router call; has `classify`. - [core/llm-server/eval/EvalAdapter.md](core/llm-server/eval/EvalAdapter.md): wires eval runs to AgentRunner or the chat bridge; has `makeAgentRun`, `makeBridgeAgentRun`. - [core/llm-server/server/ServerLauncher.md](core/llm-server/server/ServerLauncher.md): starts the local llama-server from saved defaults; has `shared`, `resolveAndStart`. - [core/llm-server/server/FitTester.md](core/llm-server/server/FitTester.md): measured VRAM, RAM and speed fit test across contexts and KV modes; has `shared`, `run`. - [core/llm-server/server/launcher/LaunchRun.md](core/llm-server/server/launcher/LaunchRun.md): one llama-server start, step by step; has `execute`. - [core/llm-server/server/launcher/LaunchContext.md](core/llm-server/server/launcher/LaunchContext.md): per-start state (OpenVINO device included) plus the planning call; has `plan`, `syncQueue`, `succeed`. - [core/llm-server/server/launcher/LaunchPreflight.md](core/llm-server/server/launcher/LaunchPreflight.md): pre-launch diagnostics, runtime, flags, model and key checks; has `resolve`. - [core/llm-server/server/launcher/ModelRuntimeMatch.md](core/llm-server/server/launcher/ModelRuntimeMatch.md): refuses a model the chosen runtime cannot run; has `refusal`. - [core/llm-server/server/launcher/LaunchOverrides.md](core/llm-server/server/launcher/LaunchOverrides.md): planner overrides from defaults, layout and measured fit, plus the OpenVINO device (one slot on the NPU); has `apply`. - [core/llm-server/server/launcher/LlmLayoutIntent.md](core/llm-server/server/launcher/LlmLayoutIntent.md): what the placement layout says about the LLM; has `read`. - [core/llm-server/server/launcher/PeerGpuBorrow.md](core/llm-server/server/launcher/PeerGpuBorrow.md): borrows peer GPUs over RPC when needed; has `borrow`, `release`. - [core/llm-server/server/launcher/LlmPlacement.md](core/llm-server/server/launcher/LlmPlacement.md): LLM VRAM reserve and device narrowing; has `place`, `placeForRescue`. - [core/llm-server/server/launcher/LaunchFinalizer.md](core/llm-server/server/launcher/LaunchFinalizer.md): auth key, pin, runtime env (OpenVINO device), health budget and split for a launch; has `finalize`. - [core/llm-server/server/launcher/FlagRescue.md](core/llm-server/server/launcher/FlagRescue.md): one-shot retry after an unknown flag; has `attempt`. - [core/llm-server/server/launcher/OomRescue.md](core/llm-server/server/launcher/OomRescue.md): one-shot retry after running out of memory; has `attempt`. - [core/llm-server/server/launcher/OpenVinoRescue.md](core/llm-server/server/launcher/OpenVinoRescue.md): one-shot retry on the CPU device after OpenVINO could not open its GPU or NPU; has `attempt`, `note`. - [core/llm-server/server/fit-test/FitComboRunner.md](core/llm-server/server/fit-test/FitComboRunner.md): measures one fit-test combination; has `run`. - [core/llm-server/server/fit-test/FitComboLadder.md](core/llm-server/server/fit-test/FitComboLadder.md): fit-test combinations and the depth probe pick; has `build`, `pickDepthProbe`. - [core/llm-server/server/fit-test/FitGenerationClient.md](core/llm-server/server/fit-test/FitGenerationClient.md): timed completions and token counts for fit tests; has `generateAndTime`, `countTokens`. - [core/llm-server/server/fit-test/DepthProbe.md](core/llm-server/server/fit-test/DepthProbe.md): long-context generation speed probe; has `measure`. - [core/llm-server/server/fit-test/VramFootprint.md](core/llm-server/server/fit-test/VramFootprint.md): VRAM and RAM peak during a fit test; has `captureBaseline`, `track`, `resolve`. - [core/llm-server/server/fit-test/FitRunSignals.md](core/llm-server/server/fit-test/FitRunSignals.md): fit-test cancel and progress hooks; has `cancelled`, `emit`. - [core/llm-server/server/fit-test/CancellableDelay.md](core/llm-server/server/fit-test/CancellableDelay.md): cancel-aware sleep and settle wait; has `sleep`, `until`. - [core/llm-server/LLMServerService.md](core/llm-server/LLMServerService.md): owns the local LLM feature (settings, defaults, supervisor, watchdog, RAM pin, router, caches, pinned tab); has `ensureRunning`, `getEffectiveContext`, `getDefaults`, `setDefaults`, `computeLocalProviderEntry`, `setEnabled`, `shutdown`. - [core/llm-server/service/SettingsMapStore.md](core/llm-server/service/SettingsMapStore.md): base for one object map under one settings key; has `all`. - [core/llm-server/service/ModelTextOverrides.md](core/llm-server/service/ModelTextOverrides.md): base for per-model text overrides; has `get`, `set`. - [core/llm-server/service/LlmModelDisplayNames.md](core/llm-server/service/LlmModelDisplayNames.md): display-name overrides for local models; has `all`, `set`, `resolve`. - [core/llm-server/service/ModelLaunchFlags.md](core/llm-server/service/ModelLaunchFlags.md): extra llama-server flags per model; has `resolve`, `setForModel`. - [core/llm-server/service/ModelOverrideStore.md](core/llm-server/service/ModelOverrideStore.md): sparse per-model setting overrides keyed by chat model ref; has `get`, `set`. - [core/llm-server/service/ModelProfileStore.md](core/llm-server/service/ModelProfileStore.md): named model profiles (a base model plus request overrides); has `listFor`, `create`, `update`. - [core/llm-server/service/ModelSettings.md](core/llm-server/service/ModelSettings.md): the global, model and profile setting layers; resolves profile refs to their base model and backs the Tune panel; has `resolve`, `baseRef`, `view`, `setModel`, `saveProfile`, `deleteProfile`, `launchDefaults`. - [core/llm-server/service/ModelResultStore.md](core/llm-server/service/ModelResultStore.md): base for saved per-model run results; has `get`, `save`. - [core/llm-server/service/FitResultStore.md](core/llm-server/service/FitResultStore.md): saved fit-test results; has `all`, `get`, `save`. - [core/llm-server/service/GambitResultStore.md](core/llm-server/service/GambitResultStore.md): saved model compatibility results; has `all`, `get`, `save`. - [core/llm-server/service/KeyedSettings.md](core/llm-server/service/KeyedSettings.md): base for settings spread over several keys. - [core/llm-server/service/LlmServerSettings.md](core/llm-server/service/LlmServerSettings.md): local LLM feature settings; has `isEnabled`, `setEnabled`, `getModelsDirConfig`, `setAutoUnloadMs`. - [core/llm-server/service/LlmUiState.md](core/llm-server/service/LlmUiState.md): LLM tab UI state; has `getMode`, `setMode`, `takeChatIntent`. - [core/llm-server/service/LlmDefaults.md](core/llm-server/service/LlmDefaults.md): saved launch defaults, OpenVINO device included; has `get`, `set`. - [core/llm-server/service/NvidiaSmiPathSettings.md](core/llm-server/service/NvidiaSmiPathSettings.md): saved nvidia-smi path; has `getSavedPath`, `setSavedPath`. - [core/llm-server/service/LlmDiagnosticsCache.md](core/llm-server/service/LlmDiagnosticsCache.md): cached hardware diagnostics, re-probed when the schema stamp is old; has `getCached`, `setCached`, `ensure`. - [core/llm-server/service/LocalRequestTracker.md](core/llm-server/service/LocalRequestTracker.md): in-flight local request counts; has `start`, `end`, `isBusy`. - [core/llm-server/service/LocalServerStarter.md](core/llm-server/service/LocalServerStarter.md): on-demand start and vision restart; has `ensureRunning`, `needsVisionLoad`. - [core/llm-server/service/RunningModelCaps.md](core/llm-server/service/RunningModelCaps.md): the running model's capabilities; has `get`, `runningThinking`. - [core/llm-server/service/LocalProviderEntry.md](core/llm-server/service/LocalProviderEntry.md): the local provider row; has `compute`, `queueKey`. - [core/llm-server/service/InstalledChatModels.md](core/llm-server/service/InstalledChatModels.md): installed chat models; has `list`. - [core/llm-server/service/PinnedLlmTab.md](core/llm-server/service/PinnedLlmTab.md): the pinned LLM tab; has `ensure`, `remove`, `send`. - [core/llm-server/service/LlmSupervisorHooks.md](core/llm-server/service/LlmSupervisorHooks.md): wires the LLM supervisor to the VRAM ledger, watchdog and caches; has `wire`. - [core/llm-server/chat/TriggerRunner.md](core/llm-server/chat/TriggerRunner.md): runs trigger fires with gating, a per-trigger queue and one run at a time; has `fire`, `deliver`, `runInline`, `simulate`, `replay`, `approve`, `stop`. - [core/llm-server/chat/ArtifactTaskScheduler.md](core/llm-server/chat/ArtifactTaskScheduler.md): recurring runs that refresh a live widget's data; has `start`, `stop`, `tick`, `runNow`. - [core/llm-server/chat/ScheduledTaskScheduler.md](core/llm-server/chat/ScheduledTaskScheduler.md): recurring chat scheduled tasks; has `start`, `stop`, `tick`, `runNow`, `runInline`. - [core/llm-server/chat/schedulers/IntervalTaskScheduler.md](core/llm-server/chat/schedulers/IntervalTaskScheduler.md): base for tick-driven task schedulers; has `start`, `stop`, `tick`, `runNow`. - [core/llm-server/chat/schedulers/CapturedBridgeRun.md](core/llm-server/chat/schedulers/CapturedBridgeRun.md): one captured background bridge run with a pausable timeout; has `run`, `pauseTimeout`, `resumeTimeout`. - [core/llm-server/chat/schedulers/RunTranscript.md](core/llm-server/chat/schedulers/RunTranscript.md): hidden transcript of a background run and its retention; has `open`, `addUser`, `addAssistant`, `prune`. - [core/llm-server/chat/schedulers/SetupChatToolPolicy.md](core/llm-server/chat/schedulers/SetupChatToolPolicy.md): a setup conversation's run tools; has `denySet`, `runConfig`, `describe`. - [core/llm-server/chat/trigger-runner/TriggerRunExecution.md](core/llm-server/chat/trigger-runner/TriggerRunExecution.md): one trigger run from start to finish; has `execute`. - [core/llm-server/chat/trigger-runner/PendingTriggerWork.md](core/llm-server/chat/trigger-runner/PendingTriggerWork.md): base for timer-held trigger work; has `has`, `dropAll`. - [core/llm-server/chat/trigger-runner/TriggerBatchWindows.md](core/llm-server/chat/trigger-runner/TriggerBatchWindows.md): trigger batch windows; has `add`, `depth`. - [core/llm-server/chat/trigger-runner/QuietHoursHold.md](core/llm-server/chat/trigger-runner/QuietHoursHold.md): holds events until quiet hours end; has `add`, `status`. - [core/llm-server/chat/trigger-runner/TriggerRetryScheduler.md](core/llm-server/chat/trigger-runner/TriggerRetryScheduler.md): jittered retries of failed trigger runs; has `schedule`, `pending`. - [core/llm-server/chat/trigger-runner/TriggerApprovalHold.md](core/llm-server/chat/trigger-runner/TriggerApprovalHold.md): parks trigger runs on human approval; has `onRequest`, `pending`, `approve`. - [core/llm-server/chat/trigger-runner/TriggerAdmission.md](core/llm-server/chat/trigger-runner/TriggerAdmission.md): loop guard, filter and cooldown gates; has `refusal`, `markAccepted`. - [core/llm-server/chat/trigger-runner/TriggerDriftCheck.md](core/llm-server/chat/trigger-runner/TriggerDriftCheck.md): records and notifies payload drift; has `note`. - [core/llm-server/chat/trigger-runner/TriggerFailureNotifier.md](core/llm-server/chat/trigger-runner/TriggerFailureNotifier.md): failure streak and auto-pause notifications; has `afterFire`. - [core/llm-server/chat/trigger-runner/TriggerRunConfig.md](core/llm-server/chat/trigger-runner/TriggerRunConfig.md): run model and tools for a trigger; has `resolve`. - [core/llm-server/chat/trigger-runner/TriggerRunTools.md](core/llm-server/chat/trigger-runner/TriggerRunTools.md): run-scoped trigger tools and allow-list; has `build`, `allowList`. - [core/llm-server/chat/trigger-runner/SaveMemoryTool.md](core/llm-server/chat/trigger-runner/SaveMemoryTool.md): the save_memory trigger tool; has `definition`. - [core/llm-server/chat/trigger-runner/RespondToWebhookTool.md](core/llm-server/chat/trigger-runner/RespondToWebhookTool.md): the respond_to_webhook trigger tool; has `definition`. - [core/llm-server/chat/trigger-runner/TriggerResultShape.md](core/llm-server/chat/trigger-runner/TriggerResultShape.md): checks a trigger result against its expected shape; has `resultJson`, `error`. - [core/llm-server/chat/trigger-runner/TriggerSampleEvent.md](core/llm-server/chat/trigger-runner/TriggerSampleEvent.md): turns a sample into a trigger event; has `for`. - [core/llm-server/chat/trigger-runner/TriggerRunPrompt.md](core/llm-server/chat/trigger-runner/TriggerRunPrompt.md): builds the trigger run system prompt; has `system`, `memoryBlock`. - [core/llm-server/chat/trigger-runner/TriggerDeliveryLogger.md](core/llm-server/chat/trigger-runner/TriggerDeliveryLogger.md): delivery-log writer that never throws; has `log`. - [core/llm-server/chat/trigger-runner/HeldGate.md](core/llm-server/chat/trigger-runner/HeldGate.md): a run's hold on the background run gate; has `acquire`, `release`, `reacquire`. - [core/llm-server/chat/trigger-runner/TriggerRunQueue.md](core/llm-server/chat/trigger-runner/TriggerRunQueue.md): global trigger queue with a per-trigger bound; has `push`, `shift`, `depth`. - [core/llm-server/ui/SharedUiLibrary.md](core/llm-server/ui/SharedUiLibrary.md): `core/llm-server/ui/js/`, `core/llm-server/ui/resonant.js`, `core/shell/ui/luma-modal.js`, `extensions/ui-kit/ui/`, `extensions/ext-ui.js`. - [core/llm-server/ui/js/artifact-data-client.md](core/llm-server/ui/js/artifact-data-client.md): classic-script exception: sets `window.LumaArtifactData = { create }` for the standalone live-artifact page that [LiveModuleDocument](core/llm-server/chat/artifacts/LiveModuleDocument.md) writes. - [core/llm-server/ui/js/artifacts/ArtifactDataCache.md](core/llm-server/ui/js/artifacts/ArtifactDataCache.md): the localStorage write-through cache behind an [ArtifactDataStore](core/llm-server/ui/js/artifacts/ArtifactDataStore.md), so reads paint instantly and survive short offline windows. - [core/llm-server/ui/js/artifacts/ArtifactDataStore.md](core/llm-server/ui/js/artifacts/ArtifactDataStore.md): the `store` a live artifact module receives: saved data for one artifact chain. - [core/llm-server/ui/js/artifacts/ArtifactDataTransport.md](core/llm-server/ui/js/artifacts/ArtifactDataTransport.md): base class (the interface) for how an [ArtifactDataStore](core/llm-server/ui/js/artifacts/ArtifactDataStore.md) talks to the host. - [core/llm-server/ui/js/artifacts/HttpDataTransport.md](core/llm-server/ui/js/artifacts/HttpDataTransport.md): [ArtifactDataTransport](core/llm-server/ui/js/artifacts/ArtifactDataTransport.md) over HTTP, for the pop-out page, the sharing PWA and read-only `/share` views. - [core/llm-server/ui/js/artifacts/IpcDataTransport.md](core/llm-server/ui/js/artifacts/IpcDataTransport.md): [ArtifactDataTransport](core/llm-server/ui/js/artifacts/ArtifactDataTransport.md) over the preload's `artifactData` surface (`{ all, mutate, onChanged }`), used by the LLM tab and the Dashboard. - [core/llm-server/ui/js/dialogs/Dialogs.md](core/llm-server/ui/js/dialogs/Dialogs.md): promise-returning alert, confirm and prompt for module code. - [core/llm-server/ui/js/dom/Clipboard.md](core/llm-server/ui/js/dom/Clipboard.md): copies text to the clipboard with a fallback. - [core/llm-server/ui/js/dom/Dom.md](core/llm-server/ui/js/dom/Dom.md): the one element factory and id lookup for renderer code. - [core/llm-server/ui/js/dom/HtmlPatcher.md](core/llm-server/ui/js/dom/HtmlPatcher.md): writes new HTML into an element keeping the leading nodes that did not change; has `apply`. - [core/llm-server/ui/js/format/ByteFormatter.md](core/llm-server/ui/js/format/ByteFormatter.md): the renderer's byte ladder and gigabyte card label. - [core/llm-server/ui/js/format/HtmlEscaper.md](core/llm-server/ui/js/format/HtmlEscaper.md): the renderer's HTML escapers. - [core/llm-server/ui/js/format/TransferText.md](core/llm-server/ui/js/format/TransferText.md): download progress wording shared by every setup surface. - [core/llm-server/ui/js/live/ChartLoader.md](core/llm-server/ui/js/live/ChartLoader.md): loads the vendored Chart.js once, on the first live module that asks for it. - [core/llm-server/ui/js/live/LiveModuleMounter.md](core/llm-server/ui/js/live/LiveModuleMounter.md): the one mounter for live (interactive) artifacts in the chat, the web PWA and the Dashboard, so the three cannot drift. - [core/llm-server/ui/js/live/LiveModuleSource.md](core/llm-server/ui/js/live/LiveModuleSource.md): reads a live module's JS for the names the mount injects. - [core/llm-server/ui/js/live/LumaBridge.md](core/llm-server/ui/js/live/LumaBridge.md): builds the `luma` page-API object a live module receives (`fetchPage`, `openTab`, `ext(extensionId)` for extension APIs such as the Hub). - [core/llm-server/ui/js/markdown/AttachmentParser.md](core/llm-server/ui/js/markdown/AttachmentParser.md): splits the attachment blocks the chat composer splices onto the top of a user message back out, so renderers (chat bubble, web client, share viewer) can show each file as a card. - [core/llm-server/ui/js/markdown/ChartWidget.md](core/llm-server/ui/js/markdown/ChartWidget.md): the ```chart fence: a JSON spec rendered as an inline SVG column or line chart (validated palette, tooltips, legend from two series, a data table). - [core/llm-server/ui/js/markdown/CodeHighlighter.md](core/llm-server/ui/js/markdown/CodeHighlighter.md): the chat's compact, language-agnostic code highlighter. - [core/llm-server/ui/js/markdown/MarkdownBlocks.md](core/llm-server/ui/js/markdown/MarkdownBlocks.md): the block pass of [MarkdownRenderer](core/llm-server/ui/js/markdown/MarkdownRenderer.md). - [core/llm-server/ui/js/markdown/MarkdownInline.md](core/llm-server/ui/js/markdown/MarkdownInline.md): the chat markdown's inline rules, applied to text that is already escaped. - [core/llm-server/ui/js/markdown/MarkdownRenderer.md](core/llm-server/ui/js/markdown/MarkdownRenderer.md): the chat's bespoke markdown renderer, used by the chat, the web client, the Dashboard, On Demand and the share viewer. - [core/llm-server/ui/js/markdown/StatsWidget.md](core/llm-server/ui/js/markdown/StatsWidget.md): the ```stats fence: up to four stat tiles (label, value, a delta with a direction arrow). - [core/llm-server/ui/js/markdown/StreamingMarkdown.md](core/llm-server/ui/js/markdown/StreamingMarkdown.md): closes the markdown a streaming reply has left open (code fence, inline code, bold, a link still arriving) so live renders never flash raw syntax. - [core/llm-server/ui/js/monaco/MonacoDiagnostics.md](core/llm-server/ui/js/monaco/MonacoDiagnostics.md): turns on Monaco's built-in squiggles, tuned for standalone snippets. - [core/llm-server/ui/js/monaco/MonacoLanguages.md](core/llm-server/ui/js/monaco/MonacoLanguages.md): maps an artifact's language tag or a file name to a Monaco language id. - [core/llm-server/ui/js/monaco/MonacoLoader.md](core/llm-server/ui/js/monaco/MonacoLoader.md): lazily loads the vendored Monaco editor (about 5 MB parsed) once, on first use. - [core/llm-server/ui/js/monaco/MonacoThemes.md](core/llm-server/ui/js/monaco/MonacoThemes.md): the LumaBrowser editor theme. - [core/llm-server/ui/js/resonant/ResonantRuntime.md](core/llm-server/ui/js/resonant/ResonantRuntime.md): the ES-module door to ResonantJs ([resonant.js](core/llm-server/ui/resonant.md), a classic vendor script the page loads before its module entry). - [core/llm-server/ui/js/resonant/ResonantTemplates.md](core/llm-server/ui/js/resonant/ResonantTemplates.md): the small, deliberate set of ResonantJs templates shared across LLM-tab sections. - [core/llm-server/ui/js/setup/AutoImageChoice.md](core/llm-server/ui/js/setup/AutoImageChoice.md): the plan screen's one image choice, shared by both setup front doors. - [core/llm-server/ui/js/setup/AutoSetup.md](core/llm-server/ui/js/setup/AutoSetup.md): the full Automatic Local Setup [pipeline](core/llm-server/ui/js/setup/SetupPipeline.md), from a planner plan. - [core/llm-server/ui/js/setup/AutoSetupProgress.md](core/llm-server/ui/js/setup/AutoSetupProgress.md): one progress bar across every download leg of an [AutoSetup](core/llm-server/ui/js/setup/AutoSetup.md) plan. - [core/llm-server/ui/js/setup/ExistingImagePlan.md](core/llm-server/ui/js/setup/ExistingImagePlan.md): lets an Automatic Setup plan link an image checkpoint the user already has (ComfyUI, Forge, ...) instead of downloading the catalog model. - [core/llm-server/ui/js/setup/ExistingLibraryView.md](core/llm-server/ui/js/setup/ExistingLibraryView.md): shapes a scan of models the user already has for both setup front doors. - [core/llm-server/ui/js/setup/FoldMemory.md](core/llm-server/ui/js/setup/FoldMemory.md): remembers which collapsible Setup sections the user opened. - [core/llm-server/ui/js/setup/HardwareText.md](core/llm-server/ui/js/setup/HardwareText.md): the one hardware line every onboarding screen shows. - [core/llm-server/ui/js/setup/ImageModelBudget.md](core/llm-server/ui/js/setup/ImageModelBudget.md): the setup wizards' quality-first image model pick against ONE card's memory. - [core/llm-server/ui/js/setup/ImageRuntimePicker.md](core/llm-server/ui/js/setup/ImageRuntimePicker.md): picks the image runtime for setup. - [core/llm-server/ui/js/setup/ImageSetup.md](core/llm-server/ui/js/setup/ImageSetup.md): the image-server [setup pipeline](core/llm-server/ui/js/setup/SetupPipeline.md). - [core/llm-server/ui/js/setup/ImportSetup.md](core/llm-server/ui/js/setup/ImportSetup.md): the [setup pipeline](core/llm-server/ui/js/setup/SetupPipeline.md) that adopts an LLM the user already has (LM Studio, Hugging Face cache, Ollama) instead of downloading one. - [core/llm-server/ui/js/setup/LlmSetup.md](core/llm-server/ui/js/setup/LlmSetup.md): the local-LLM [setup pipeline](core/llm-server/ui/js/setup/SetupPipeline.md): runtime, resumable model download, defaults, server start. - [core/llm-server/ui/js/setup/MusicSetup.md](core/llm-server/ui/js/setup/MusicSetup.md): the music-server [setup pipeline](core/llm-server/ui/js/setup/SetupPipeline.md). - [core/llm-server/ui/js/setup/QuantText.md](core/llm-server/ui/js/setup/QuantText.md): plain-language glosses for GGUF quant tags (tooltips on the wizards' quant badges). - [core/llm-server/ui/js/setup/RamPin.md](core/llm-server/ui/js/setup/RamPin.md): turns on RAM pinning for the chat and image servers after a singularity setup. - [core/llm-server/ui/js/setup/RuntimeEnsurer.md](core/llm-server/ui/js/setup/RuntimeEnsurer.md): makes sure the LLM inference runtime is installed. - [core/llm-server/ui/js/setup/RuntimeInstallModal.md](core/llm-server/ui/js/setup/RuntimeInstallModal.md): the runtime install or update chooser shared by the LLM and image runtime cards. - [core/llm-server/ui/js/setup/SegmentedPicker.md](core/llm-server/ui/js/setup/SegmentedPicker.md): a segmented pill picker backed by a hidden input, for short closed sets (2 to 6 values). - [core/llm-server/ui/js/setup/ServerStarter.md](core/llm-server/ui/js/setup/ServerStarter.md): starts the LLM server idempotently. - [core/llm-server/ui/js/setup/SetupHooks.md](core/llm-server/ui/js/setup/SetupHooks.md): the view callbacks a setup pipeline reports through. - [core/llm-server/ui/js/setup/SetupPipeline.md](core/llm-server/ui/js/setup/SetupPipeline.md): base class for the setup pipelines that both onboarding surfaces drive (the LLM tab's Easy Setup and the first-run SetupWizard): [LlmSetup](core/llm-server/ui/js/setup/LlmSetup.md), [ImportSetup](core/llm-server/ui/js/setup/ImportSetup.md), [ImageSetup](core/llm-server/ui/js/setup/ImageSetup.md), [MusicSetup](core/llm-server/ui/js/setup/MusicSetup.md) and [AutoSetup](core/llm-server/ui/js/setup/AutoSetup.md). - [core/llm-server/ui/js/setup/SetupProgressEvents.md](core/llm-server/ui/js/setup/SetupProgressEvents.md): turns runtime-install and model-download events into progress-bar updates with the same wording on every setup leg. - [core/llm-server/ui/js/setup/SetupQuestions.md](core/llm-server/ui/js/setup/SetupQuestions.md): the guided wizards' three questions, held once. - [core/llm-server/ui/js/setup/SpeedSimulator.md](core/llm-server/ui/js/setup/SpeedSimulator.md): the wizards' live "speed tolerance" preview. - [core/llm-server/ui/js/setup/SystemLibraries.md](core/llm-server/ui/js/setup/SystemLibraries.md): the Linux system-library preflight, run before any download. - [core/llm-server/ui/resonant.md](core/llm-server/ui/resonant.md): resonantJs, the owner's reactive templating framework, copied byte for byte from legacy as a vendor file (allowed exception: not one class per file, not a module, and it keeps six em-dashes in its comments because the owner's rule is "copy it unchanged"). - [core/llm-server/ui/js/image-setup/CatalogModelRow.md](core/llm-server/ui/js/image-setup/CatalogModelRow.md): `CatalogModelRow.view(model, installed, activeDl, vramBytes)` builds the mlRow view of an image catalog model: category and NC pills, the graphics-memory line, the quant picker (when downloadable and idle), the size chip of the selected quant, and Auto-download, Download (NC), a disabled Download while another runs, or Cancel on the downloading row (which opens to show [DownloadDetail](core/llm-server/ui/js/image-setup/DownloadDetail.md)). - [core/llm-server/ui/js/image-setup/DownloadDetail.md](core/llm-server/ui/js/image-setup/DownloadDetail.md): `DownloadDetail.html(dl)`: the in-row progress block of the downloading catalog row ("File n of m · role", bar, percent and bytes, or "starting…"). - [core/llm-server/ui/js/image-setup/HfRepoUrl.md](core/llm-server/ui/js/image-setup/HfRepoUrl.md): `HfRepoUrl.isRepoUrl(url)`: true for a huggingface.co owner/repo page URL without a `/resolve/` or `/blob/` segment. - [core/llm-server/ui/js/image-setup/ImageDefaultsCard.md](core/llm-server/ui/js/image-setup/ImageDefaultsCard.md): the Image Defaults card: enable switch, runtime and per-job model selectors (generation, edit, video), the "More settings" fold (RAM pin, 15-minute auto-unload) and Start or Stop server. - [core/llm-server/ui/js/image-setup/ImageDefaultsMarkup.md](core/llm-server/ui/js/image-setup/ImageDefaultsMarkup.md): markup for the Image Defaults card. - [core/llm-server/ui/js/image-setup/ImageModal.md](core/llm-server/ui/js/image-setup/ImageModal.md): base for the Image Setup's hand-rolled modals: one overlay at a time, closed by Escape, the surround or `[data-act="close"]`, with the footer error line. - [core/llm-server/ui/js/image-setup/ImageModelActions.md](core/llm-server/ui/js/image-setup/ImageModelActions.md): routes the image model rows' buttons: `dl` and `dl-nc` (license confirm first; sends the picked quant), `dl-cancel`, `pin` (the default of the button's category), `move` (set the stored kind), `loras`, `update-file` (companion file swap) and `uninstall` (confirm first). - [core/llm-server/ui/js/image-setup/ImageModelsCard.md](core/llm-server/ui/js/image-setup/ImageModelsCard.md): the Image Models card: a one-time scaffold of two remembered folds holding two STABLE [ModelList](core/llm-server/ui/js/models/ModelList.md) mounts (`mlImgInstalled` with the image library scan, `mlImgCatalog` with the quant pickers), plus repainted chrome: the download error banner, the models-folder controls, the empty note, "+ Import custom model" and the fine-tune hint. - [core/llm-server/ui/js/image-setup/ImageRuntimeRow.md](core/llm-server/ui/js/image-setup/ImageRuntimeRow.md): markup for one collapsible image runtime row: status badges ("ready" needs the hardware AND an installed binary or a prebuilt asset for this host), the primary Install or Update (hidden while installing, shown again after a failure), details with version, path and secondary actions, and the progress strip. - [core/llm-server/ui/js/image-setup/ImageRuntimesCard.md](core/llm-server/ui/js/image-setup/ImageRuntimesCard.md): the Image Runtimes card: installed runtimes above a Catalogue fold (open while one of its rows installs or nothing is installed), Install or Update through [RuntimeInstallModal](core/llm-server/ui/js/setup/RuntimeInstallModal.md) (prebuilt download or locate), Uninstall, Clear registration, Relocate, live install progress and update badges. - [core/llm-server/ui/js/image-setup/ImageSetupPanel.md](core/llm-server/ui/js/image-setup/ImageSetupPanel.md): the Image section of the LLM tab's Setup view (cards `cardImageDefaults`, `cardImageRuntimes`, `cardImageModels`). - [core/llm-server/ui/js/image-setup/ImageSetupStore.md](core/llm-server/ui/js/image-setup/ImageSetupStore.md): the Image Setup panel's last-known state and its best-effort loaders: a failed or missing IPC leaves a safe empty value. - [core/llm-server/ui/js/image-setup/ImportModelMarkup.md](core/llm-server/ui/js/image-setup/ImportModelMarkup.md): `ImportModelMarkup.HTML`: the static markup of the Import custom image model modal. - [core/llm-server/ui/js/image-setup/ImportModelModal.md](core/llm-server/ui/js/image-setup/ImportModelModal.md): "Import custom image model": from a Hugging Face model page (recipe resolver, no name needed), a direct file URL or a local file, with the base architecture, a prompt style for bases that have profiles, and the Qwen and Anima notes. - [core/llm-server/ui/js/image-setup/InstalledModelRow.md](core/llm-server/ui/js/image-setup/InstalledModelRow.md): `InstalledModelRow.view(model, defaults, downloading)` builds the mlRow view of an installed image model: category pill ("generation + edit" for unified weights), default badges, file-update pills, NC and imported pills, and the actions Pin, Pin as edit default (unified only), Update , Move to (never Edit for a unified model), LoRAs / speed and Remove. - [core/llm-server/ui/js/image-setup/LoraBaseNote.md](core/llm-server/ui/js/image-setup/LoraBaseNote.md): `LoraBaseNote.describe(row, modelFamily)` returns `{ text, warn }`: "Trained for X.", a mismatch warning when the LoRA's families exclude the model's, and a LyCORIS LoKr/LoHa hint. - [core/llm-server/ui/js/image-setup/LoraCuratedList.md](core/llm-server/ui/js/image-setup/LoraCuratedList.md): curated speed LoRAs for the model's family (`loraCatalog`): Download (then select it and pre-check the preset), In library, or for Wan pairs Download & attach / Attach pair (both halves with the entry's preset, then close). - [core/llm-server/ui/js/image-setup/LoraModal.md](core/llm-server/ui/js/image-setup/LoraModal.md): "LoRAs / speed" for one installed image model: pick a LoRA from the shared library (annotated with its detected base), set its weight and the distilled speed preset, import one, or download a curated one; Save writes the manifest (None removes it). - [core/llm-server/ui/js/image-setup/LoraModalMarkup.md](core/llm-server/ui/js/image-setup/LoraModalMarkup.md): `LoraModalMarkup.html(label, hasPair)`: the LoRA modal markup. - [core/llm-server/ui/js/image-setup/ModelsDirControls.md](core/llm-server/ui/js/image-setup/ModelsDirControls.md): the image models-folder controls, mirroring the LLM "Models Directory" card: a path field that commits on change (Enter blurs), Browse, and "Use default" while a custom path is set. - [core/llm-server/ui/js/image-setup/ProgressPercent.md](core/llm-server/ui/js/image-setup/ProgressPercent.md): `ProgressPercent.of(received, total)`: the whole-number percent (capped at 100), or null while the total is unknown. - [core/llm-server/ui/js/image-setup/QuantPicker.md](core/llm-server/ui/js/image-setup/QuantPicker.md): the image catalog's quant picker over `files.diffusion.quants`. - [core/llm-server/ui/js/image-setup/RamPinHint.md](core/llm-server/ui/js/image-setup/RamPinHint.md): the caption on the image "RAM pin" row: pin progress, the pinned size, or the failure. - [core/llm-server/ui/js/image-setup/RuntimeProgressMarkup.md](core/llm-server/ui/js/image-setup/RuntimeProgressMarkup.md): the always-present progress strip of an image runtime row: an idle strip, a phase label with bytes and a bar (indeterminate while the total is unknown), or the failure line. - [core/llm-server/ui/js/image-setup/RuntimeUpdateBadge.md](core/llm-server/ui/js/image-setup/RuntimeUpdateBadge.md): `RuntimeUpdateBadge.apply(body, id, info)` applies a cached upstream-update result to one runtime row: the primary button reads "Update" with a version title, and an "update: vX" pill joins the head when a newer release exists. - [core/llm-server/ui/js/music-setup/AiTextCleaner.md](core/llm-server/ui/js/music-setup/AiTextCleaner.md): strips think blocks, code fences and surrounding quotes from a drafted field: `AiTextCleaner.clean(text)`, `AiTextCleaner.firstLine(text, maxChars)`. - [core/llm-server/ui/js/music-setup/MusicAiAssist.md](core/llm-server/ui/js/music-setup/MusicAiAssist.md): attaches the shared AI-fill buttons to the lyrics and style fields through the chat-extension `attachAssist` contract; this class only builds the requests (system prompt, user parts, temperature 0.9 / 0.8, 300 s timeout, noThink). - [core/llm-server/ui/js/music-setup/MusicDefaultsCard.md](core/llm-server/ui/js/music-setup/MusicDefaultsCard.md): the Music Defaults card: enable, default model, seed (blank = random), max song length (default 300 s) and idle unload. - [core/llm-server/ui/js/music-setup/MusicInstallProgress.md](core/llm-server/ui/js/music-setup/MusicInstallProgress.md): folds the music runtime's install events into the Runtime card's progress: start, resolved, download (bytes), extract (installing, keeping the last pip or uv output line), error (failed with message), finalize (null). - [core/llm-server/ui/js/music-setup/MusicModelsCard.md](core/llm-server/ui/js/music-setup/MusicModelsCard.md): the Music Models card: Download (disabled while another runs or without the server), progress with Cancel, or Installed with Delete (confirm first); an error stays on its row. - [core/llm-server/ui/js/music-setup/MusicRuntimeCard.md](core/llm-server/ui/js/music-setup/MusicRuntimeCard.md): paints the Music Runtime card (or the CUDA-only notice) and wires its buttons: install (shows "Preparing install…" at once), cancel, uninstall (confirm first), check for updates, stop server. - [core/llm-server/ui/js/music-setup/MusicRuntimeMarkup.md](core/llm-server/ui/js/music-setup/MusicRuntimeMarkup.md): markup for the Music Runtime card: name, version and status rows, hardware / WSL2 / GPU-in-WSL pills, the install progress, the "Rebuild needed" notice, and Install, Update to X, Reinstall, Cancel, Check for updates, Uninstall and Stop server. - [core/llm-server/ui/js/music-setup/MusicSetupPanel.md](core/llm-server/ui/js/music-setup/MusicSetupPanel.md): the Music section of the LLM tab's Setup view (cards `cardMusicRuntime`, `cardMusicModels`, `cardMusicDefaults`, `cardMusicTry`) for the managed SGLang-Omni server (WSL2-wrapped on Windows). - [core/llm-server/ui/js/music-setup/MusicTryCard.md](core/llm-server/ui/js/music-setup/MusicTryCard.md): the Music "Try it" card: lyrics and style with AI drafting, Generate and Cancel, the status line ("Loading model…", "Switching model…", "Composing… Ns", "Failed: …") and an audio player for the returned WAV. - [core/llm-server/ui/js/grounding-setup/GroundingMarkup.md](core/llm-server/ui/js/grounding-setup/GroundingMarkup.md): markup for the Visual grounding card. - [core/llm-server/ui/js/grounding-setup/GroundingSetupCard.md](core/llm-server/ui/js/grounding-setup/GroundingSetupCard.md): the "Visual grounding" card of the LLM tab settings. - [core/llm-server/ui/js/wizard/AutoPlanView.md](core/llm-server/ui/js/wizard/AutoPlanView.md): automatic Setup's plan screen: calls `planAutoSetup({ wantImage, wantMusic })` once, then shows the picked chat model, the download line, the planner's reasons, the detected hardware, the image-checkpoint choice ([AutoImageChoice](core/llm-server/ui/js/setup/AutoImageChoice.md), one scan per plan, undoable through `basePlan`), models already on disk, the storage line, "Set it up" and "Customize instead". - [core/llm-server/ui/js/wizard/AutoQuestionViews.md](core/llm-server/ui/js/wizard/AutoQuestionViews.md): automatic Setup's questions: `image(wizard, body)` and `music(wizard, body)`. - [core/llm-server/ui/js/wizard/AutoResultViews.md](core/llm-server/ui/js/wizard/AutoResultViews.md): automatic Setup's progress (Pause; Cancel stops BOTH the chat and the image download), done (per-leg image and music outcome) and error views (the apt line with Copy and "Check again" for missing Linux libraries, else "Try again"; "Use guided setup instead"). - [core/llm-server/ui/js/wizard/AutoSetupFlow.md](core/llm-server/ui/js/wizard/AutoSetupFlow.md): automatic Local Setup inside Easy Setup. - [core/llm-server/ui/js/wizard/ChoiceCard.md](core/llm-server/ui/js/wizard/ChoiceCard.md): one clickable Easy Setup choice card: `ChoiceCard.button(className, html, onClick)`, `ChoiceCard.titleDesc(title, desc)` (escaped). - [core/llm-server/ui/js/wizard/EasySetupWizard.md](core/llm-server/ui/js/wizard/EasySetupWizard.md): "Easy Setup", the LLM tab's onboarding overlay over the Setup surface. - [core/llm-server/ui/js/wizard/ExistingLibraries.md](core/llm-server/ui/js/wizard/ExistingLibraries.md): `ExistingLibraries.mount(wizard, host, recFor)`: lists chat models already on this machine (open as soon as one is found); picking one adopts it through [ImportFlow](core/llm-server/ui/js/wizard/ImportFlow.md) with the runtime and context from `recFor()`. - [core/llm-server/ui/js/wizard/ExistingModelsMarkup.md](core/llm-server/ui/js/wizard/ExistingModelsMarkup.md): the "You already have N models on this machine. - [core/llm-server/ui/js/wizard/GuidedSteps.md](core/llm-server/ui/js/wizard/GuidedSteps.md): the question steps: `welcome` (Automatic or Guided), `useCase`, `speed` (one live typing sim per speed, fastest last) and `context` (sample snippets). - [core/llm-server/ui/js/wizard/HfSearchPanel.md](core/llm-server/ui/js/wizard/HfSearchPanel.md): the live Hugging Face search under the recommendation: Search or Enter, results with downloads and gated marks, a repo opens its quants once (multi-part quants are refused with "paste URL"), and a pick becomes `state.override` (MLX picks carry `runtimeId: 'mlx-lm'`) with a Clear chip. - [core/llm-server/ui/js/wizard/HostPlatform.md](core/llm-server/ui/js/wizard/HostPlatform.md): renderer-side platform checks shared by the Setup panels: `isMacPlatform(nav?)` (navigator.platform; music and the music question) and `isMac(nav?)` (platform or user agent; the MLX search toggles). - [core/llm-server/ui/js/wizard/ImageSetupFlow.md](core/llm-server/ui/js/wizard/ImageSetupFlow.md): runs [ImageSetup](core/llm-server/ui/js/setup/ImageSetup.md) for the image step (runtime plus model download or a found checkpoint) and lands on its done or error view. - [core/llm-server/ui/js/wizard/ImageStep.md](core/llm-server/ui/js/wizard/ImageStep.md): step 5, optional image generation. - [core/llm-server/ui/js/wizard/ImportFlow.md](core/llm-server/ui/js/wizard/ImportFlow.md): adopts a found chat model with [ImportSetup](core/llm-server/ui/js/setup/ImportSetup.md): progress, then "is set up and running. - [core/llm-server/ui/js/wizard/LlmSetupFlow.md](core/llm-server/ui/js/wizard/LlmSetupFlow.md): the guided "Download & set up": [LlmSetup](core/llm-server/ui/js/setup/LlmSetup.md) with the recommendation, the search override and the pasted URL, Pause and Cancel, then "Local AI is ready" with Continue to image generation or Skip. - [core/llm-server/ui/js/wizard/LoadingSteps.md](core/llm-server/ui/js/wizard/LoadingSteps.md): `LoadingSteps.mount(body, steps, timers, stepMs?)`: a spinner over a step list that ticks forward every 1.8 s while a single-call IPC (plan, recommend, catalog) runs; the last step stays lit. - [core/llm-server/ui/js/wizard/PausedView.md](core/llm-server/ui/js/wizard/PausedView.md): `PausedView.show(wizard, body, resume)`: "Download paused" with Resume; the bytes stay on disk with their chunk map, so resuming (now or after a restart) continues where it stopped. - [core/llm-server/ui/js/wizard/ProgressBar.md](core/llm-server/ui/js/wizard/ProgressBar.md): `ProgressBar.set(bar, fill, fraction)`: a fraction fills the bar; null makes it indeterminate and leaves the width to the CSS keyframes. - [core/llm-server/ui/js/wizard/RecommendStep.md](core/llm-server/ui/js/wizard/RecommendStep.md): step 4: asks `recommendModel(answers)`, then shows the pick (label, quant, size, context, runtime, rationale, hardware, CPU-only or spill warning), the HF search and URL paste, models already on disk, the storage line and "Download & set up". - [core/llm-server/ui/js/wizard/StorageLine.md](core/llm-server/ui/js/wizard/StorageLine.md): `StorageLine.mount(host, needBytes, api)`: "path · N GB free · Change" under a recommendation, with a warning when the drive has under 110% of the download free. - [core/llm-server/ui/js/wizard/WizardState.md](core/llm-server/ui/js/wizard/WizardState.md): `WizardState.initial()`: the fresh state of one Easy Setup run (step, answers, rec, hw, advanced, busy, mode, the `auto` bag, `override`, `llmDone`, the `image` bag). - [core/llm-server/ui/js/wizard/WizardTimers.md](core/llm-server/ui/js/wizard/WizardTimers.md): the overlay's interval timers (speed sims, loading tickers): `add(id)`, `clear()`. - [core/llm-server/ui/js/models/LibraryScan.md](core/llm-server/ui/js/models/LibraryScan.md): "Scan other tools' libraries" under a model list: runs the existing-library scan (LM Studio, Ollama, the Hugging Face cache; ComfyUI and friends for images), offers to link each found model into the models folder (never a copy), and asks Setup to refresh. - [core/llm-server/ui/js/models/LibraryScanMarkup.md](core/llm-server/ui/js/models/LibraryScanMarkup.md): markup for LibraryScan: `head(text, withPicker)`, `results(view, skipped, text)` and `failure(error)`. - [core/llm-server/ui/js/models/ModelList.md](core/llm-server/ui/js/models/ModelList.md): the one renderer for the LLM "Models Directory" and the Image "Image Models" lists. - [core/llm-server/ui/js/models/ModelListController.md](core/llm-server/ui/js/models/ModelListController.md): the handle a section gets back from `ModelList.mount`. - [core/llm-server/ui/js/models/ModelSearch.md](core/llm-server/ui/js/models/ModelSearch.md): the "Search models" modal of the LLM Setup area: a two-pane Hugging Face GGUF (or, on a Mac, MLX) browser. - [core/llm-server/ui/js/models/ModelSearchDetail.md](core/llm-server/ui/js/models/ModelSearchDetail.md): the right pane of [ModelSearch](core/llm-server/ui/js/models/ModelSearch.md): repo facts (params, context, architecture), a quant `