# Agent Decisions

Use this category for programs where Jev supplies the decision step inside an agentic loop — tool choice, escalation, retry or stop, next-action selection.

## Submission format

```md
- [Name](URL) - Industry: one-sentence description of the Jev use case.
```

## Entries

- [Learn Jev end to end](https://github.com/harshithsunku/learn-jev-end-to-end) - Developer education: a 12-notebook Python course whose hand-rolled agent loop asks Jev a `Choice` (allow / ask / block) with `Noul` irreversibility and exfiltration checks before every tool call, sends ask verdicts to a human and fails closed on errors, and adds a `Choice` model router with a confidence fallback and a `Noul` "am I done?" gate, each measured against labeled fixtures.
- [Hermes JIT Context OS](https://github.com/wojciechwiesner/jit-context) - Coding agents: uses Jev as a sub-millisecond System 1 Epistemic Gate and Domain Router to score AST relevance, test proofs, and tool targets, cutting autonomous agent turns by 31.3% and blind file exploration by 52.6% on SWE-bench with fail-open circuit-breaker resilience.
- [Jev by Example](https://github.com/ReallyArtificial/jev-by-example) - Agent development: runnable JavaScript lessons use Jev Choice, Score, and Noul judgments for memory reconciliation, recovery proposals, and handoff checks, with explicit application policies, offline fixtures, and opt-in live calls.
- [jev-social](https://github.com/socai-io/jev-social) - Social media research: uses a Jev `Choice` at each step to select a concrete socai CLI operation and observed post or profile target on Instagram, TikTok, or LinkedIn, rejecting malformed or low-confidence decisions before execution.
- [Jev Ultrafast](https://github.com/browser-use/jev-ultrafast) - Browser automation: browser-use's ultrafast agent where Jev decides each next action and element to click, calling a language model only when text must be typed.
- [jev-agent-browser](https://github.com/forvela/jev-agent-browser) - Browser agents: a parent agent delegates bounded tasks to a Jev loop that selects typed browser actions, validates them through agent-browser, and escalates ambiguity or stuck states back to the parent.
- [pi-typesafe-jev](https://github.com/legacybridge-tech/pi-typesafe-jev) `{agent: pi, type: plugin}` - Coding agents: exposes System One judgments as five Pi tools so a model makes narrow semantic judgments while code and users keep control of thresholds, weights, and actions.
- [jev-judgment](https://github.com/HyunjunJeon/jev-judgment) - Coding agents: agent skill that sends closed coding-agent judgments to Jev so verdicts stay typed, cheap, and comparable across runs.
- [limpet](https://github.com/noplan-inc/limpet) - Coding agents: Stop hook that keeps an agent from finishing too early by judging plain-language completion rules with Jev.
- [robo-harness](https://github.com/grmkris/robo-harness) - Robotics: SO-101 arm workbench where a Jev decision runner picks bounded joint steps from typed candidate actions under a spend budget.
- [dsh-auto-mode](https://git.allen-software.com/allenh1/dsh-auto-mode) - Coding agents: DeepSeek Harness permission preset whose end-prompt step has Jev answer the open questions an agent leaves in its final message, steering them back only when a choice clears 0.6 confidence and an autonomy-safety Noul clears 0.5, and returning the turn to the human otherwise.
- [augustus](https://github.com/24601/Augustus) - Coding agents: independent skill for building and improving decision-model systems with Choice/Score/Noul, composition rules, evaluation harnesses, and bounded prompt/program optimization, using TypeSafe Jev as the default hosted exemplar.
- [yoshi](https://github.com/compozy/yoshi) `{agent: multi, type: proxy}` - Context management: proxy for Claude Code and Codex where Jev judges which conversation history is still needed before pruning.
- [pi-jev (TheoOliveira)](https://github.com/TheoOliveira/pi-jev) `{agent: pi}` - Coding agents: semantic tool routing and typed System One decisions for the Pi coding agent.
- [pi-quiet-ask](https://github.com/HyunjunJeon/pi-quiet-ask) `{agent: pi}` - Coding agents: gives the Pi agent a quiet Jev decision layer for judgments it would otherwise hand to a chat model.
- [fastbrowse](https://github.com/agent-labs-dev/fastbrowse) - Browser agents: Jev picks each action from what is on the page while an LLM reads and plans.
- [super-jev](https://github.com/Kevthetech143/super-jev) - Decision harness: turns a Jev answer into a bounded action instead of leaving the caller to interpret it.
- [jev-superpowers](https://github.com/AkashPriyadarshii/jev-superpowers) - Coding agents: software development framework for AI coding agents that hands package vetting and completion gates to Jev typed decisions.
- [Jev Browser](https://github.com/jkudish/jev-browser) - Browser automation: drives a browser with Jev deciding each step, pitched as fast and very cheap next to LLM-driven browsing.
- [pi-fast-jev-compaction](https://github.com/joelhooks/pi-fast-jev-compaction) `{agent: pi, type: plugin}` - Context management: Pi extension that keeps conversation text verbatim while pruning stale tool history with Jev, falling back to Pi's own summarization only when pruning cannot free enough room.
- [Atomic](https://github.com/bastani-inc/atomic) - Coding agent runtime: ships a first-class Jev structured-output provider so an agent's decisions come back typed, through the same decision resolver as its other providers.
- [fast-jev-compaction](https://github.com/tamaratran/fast-jev-compaction) `{agent: claude-code, type: plugin}` - Context management: Claude Code plugin that replaces the compaction summary with Jev decisions, scoring every tool call and result for whether it is still needed instead of summarizing the session.
- [fast-dev-compaction](https://github.com/leonaaardob/fast-dev-compaction) `{agent: codex, type: plugin}` - Context management: Codex port of the Jev-guided compaction idea, restoring context verbatim around a session compaction rather than summarizing it.
- [public-browser](https://github.com/Silbercue/public-browser) `{agent: multi}` - Browser control: lets Claude Code and Cursor drive a real Chrome profile, with a Jev loop deciding the actions, reporting roughly 30% fewer tokens and 25% lower cost.
- [pi-typesafe-router](https://github.com/jekozyra/pi-typesafe-router) `{agent: pi}` - Coding agents: routes Pi's work through typed Jev decisions.
- [wakegate](https://github.com/shitianfang/wakegate) - Long-running agents: before a sleeping agent's LLM is resumed on a timer or incoming event, Jev answers a `Choice` (wake, not yet, unrelated) against the agent's own sleep note, and code skips the wakeup only when wake is below 0.2 while always waking on user messages, bare timers, a skip limit, errors, and timeouts; one run passed 21 of 21 hand-written scenarios, which the README calls a smoke test rather than a benchmark.
- [BrowserClaw](https://github.com/GoldenLoaf24h/browserclaw) - Browser automation: Zero-lock, session-preserving Chrome MCP server that couples a local Jev System One semantic micro-loop (`chrome_act_toward_goal`) with an 85%+ pruned DOM tree (Shadow DOM & iframe pierced), dispatching native CDP events (`isTrusted: true`) on active logged-in sessions without focus theft.
- [jev-belay](https://github.com/valentynkit/jev-belay) `{agent: claude-code, type: plugin}` - Coding agents: Claude Code Stop hook that reads the transcript for evidence and spends one four-question Jev call only when files changed with no passing check since, failing open on any error.
- [Jev for Chrome](https://github.com/chy4pro/jev-for-chrome) `{type: extension}` - Browser automation: unofficial Chrome extension port of Jev Ultrafast where a Jev `Choice` picks the operation and DOM element each step and two `Noul` checks (goal reached, stuck) veto a premature DONE or BLOCKED, with a small text model used only when text must be typed.
- [jev-pruner](https://github.com/tamaratran/jev-pruner) `{agent: claude-code, type: plugin}` - Context management: Claude Code plugin that trims long Bash output with Jev before the model ever sees it, keeping terminal noise out of the window.
- [jev-desktop](https://github.com/yikangy873-gif/jev-desktop) `{agent: codex}` - Computer use: supplies Jev action selection inside Codex Computer Use, choosing among desktop actions rather than asking a language model at every step.
- [jev-agent-skill](https://github.com/yuyang2230/jev-agent-skill) `{agent: multi, type: plugin}` - Developer tooling: Claude Code/ZCode skill that offloads classify/route, batch-screen, score, and compliance-check judgments to Jev via OpenCode Zen's free tier, bundling a zero-dependency jev.py caller (transient-500 retry, WAF-safe UA, GBK-pipe-safe stdin) and a production Taobao-shop comment-triage pipeline that keeps raw items out of the agent context.
- [Yappy](https://yappy.biz/jev/) - Computer use: macOS voice agent that asks Jev one `Choice` per step (operation and target control) over the front window's accessibility table, executes only validated high-confidence answers, and escalates to a full LLM agent on low confidence, no-effect actions, or unknown field values; author-reported 275–690 ms per decision.
- [JevLoop (zjunlp)](https://github.com/zjunlp/JevLoop) - Agent harness: routes the loop's own judgements to Jev, where a `Choice` picks the next tool from candidates rebuilt every step, a `Score` grades the call's risk, and a `Noul` decides whether it needs authorisation, while plain code acts on the answers so a high risk score forces human authorisation that no probability can override (7.7% of wall clock with the offline judge, 79% over the hosted API).
- [JevLoop (parkavenue9639)](https://github.com/parkavenue9639/jevloop) - Agent runtimes: a Python runtime where Jev `Choice` decisions select tools and targets, uncertain decisions escalate to an LLM, and a shared guarded kernel supports isolated Docker workspaces and paired LLM-only comparisons.
- [DataJev](https://github.com/zzz1YAO/DataJev) - Data analysis agents: an LLM performs Python-based analysis while Jev reads the compressed analytical state and decides whether the agent should continue the current direction, switch to another one, verify a finding, or stop and synthesize the answer.
- [jev-mobile](https://github.com/Friedjof/jev-mobile) - Mobile control: fast structured Android control loops that route each step through Jev alongside Mobile MCP, with 35 test files.
- [GUI JEV Harness](https://github.com/ZihuaEvan/GUI_JEV) - Computer use: recursive screenshot grounding where Jev returns a `Choice` over grid-tile candidates at each level, and local probability and margin gates decide whether to descend or refuse, emitting only a raster point and bounding box and never clicking.
- [jev-compaction](https://github.com/Waxmell114514/jev-compaction) - Context management: standalone agent context compactor where Jev only scores transcript segments — kept lines stay verbatim, low scorers move to a store behind an expand() pointer instead of being deleted, and the append-only frozen prefix keeps the prompt cache valid; runnable offline demo, no API key needed.
- [Visual-JEV](https://github.com/jiangxiluning/Visual-Jev) - Multimodal models: Jev-style model built on Qwen3.5-4B that takes images directly, without first converting them to text.
- [DeepSearcher stopping-policy experiment](https://github.com/zilliztech/deep-searcher/blob/master/evaluation/jev_stopping/README.md) - Agentic search: a standalone evaluation uses Jev Noul judgments on accumulated evidence to decide whether to stop or continue within a search-round budget, comparing stopping behavior, evidence recall, and decision cost.
- [OmniJev](https://github.com/shapsider/OmniJev) - Embodied robotics: a Jev-style finite-choice interface that feeds dual-camera images and text to a self-hosted multimodal model and takes the next preset skill for a MuJoCo arm as one typed choice, where released episodes finish transfer, stack and barrier tasks in 13 decisions and 39 output tokens each, 208/208 non-audio probe requests answer correctly, unobservable inputs come back "insufficient evidence" instead of a guess, and four public benchmark pilots hold accuracy equal to a direct short answer while cutting decision latency 6.8–13.6× and total tokens 51–86% (audio input is wired in the client but rejected by the current backend).
- [neo4jev](https://github.com/jexp/neo4jev) - Graph navigation: navigates a Neo4j knowledge graph hop-by-hop using Jev Choice over candidate outgoing relationships and Noul to detect goal completion, using beam search over answer log-probabilities.
- [jev-chat](https://github.com/jev-chat/jev-chat-jarvis) - Messaging: an Android accessibility service reads the conversation in WeChat, QQ, X, or Feishu, asks Jev `Choice` over candidate replies, and fills the draft box while sending stays manual; a Windows port does the same from offline OCR of the WeChat window.
- [hermes-jev-skills](https://github.com/kerpopule/hermes-jev-skills) `{agent: multi}` - Agent runtime: a Jev-powered skill suite that decides model routing, memory, compaction, skill selection, and computer or browser use for Hermes agents, and also installs under Claude Code and Codex.
- [jev-browser-use](https://github.com/wy-coliney/jev-browser-use) `{agent: codex}` - Browser automation: lets Jev pick the click while Codex thinks and verifies, reporting 5-10x faster browser operations behind four CI-run contract tests on the bridge.
- [mobile-jev](https://github.com/droidrun/mobile-jev) - Mobile agents: puts Jev into on-device screen-aware action selection for the Droidrun loop, with a Jev Studio web app streaming live device and decision telemetry.
- [Jev-cu](https://github.com/Sac-Y/Jev-cu) - Computer use: drives a GUI through Jev decisions with a `jev-decide` script and ships a P0 case set taken from accessibility-tree snapshots of a calculator, a calendar, and the NetEase home screen.
- [SkillRanker](https://github.com/Dicklesworthstone/skillranker) `{agent: claude-code, type: cli}` - Coding agents: standalone Rust CLI that uses Jev to rank candidate skills against live session context, advising the next step through a Claude Code UserPromptSubmit hook.
- [AutoGPT](https://github.com/Significant-Gravitas/AutoGPT) - Autonomous agents: open-source autonomous agent platform featuring first-class TypeSafe Jev decision blocks for typed routing, filtering, scoring, and confidence-gated next-action dispatching.