# helmd **Armor-piercing all-in-one security-analysis plugin for DeepSeek Harness** Mount one preset · Android · Web · Native · Protocol · Malware · AI-Security ready on day one English | [中文](README.md) [![Telegram](https://img.shields.io/badge/Telegram-@helm_xD-26A5E4?style=flat&logo=telegram)](https://t.me/helm_xD) [![topic: dsh-plugin](https://img.shields.io/badge/topic-dsh--plugin-2ea44f)](https://github.com/topics/dsh-plugin) [![topic: deepseek-harness](https://img.shields.io/badge/topic-deepseek--harness-2ea44f)](https://github.com/topics/deepseek-harness) [![Node >=22.19](https://img.shields.io/badge/Node-%3E%3D22.19.0-green)](https://nodejs.org) [![pnpm 11.7.0](https://img.shields.io/badge/pnpm-11.7.0-orange)](https://pnpm.io) [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)

> For learning and research only. You must comply with local laws and regulations; you are responsible for the consequences of using this project. ## Why helmd

### Armor-piercing, all-in-one Android · Web · Native · Protocol · Malware · AI-Security converge in a single preset. Install once and every domain's tools are ready — no per-domain wiring.	### Ten bundles · zero wiring 10 `@dsh-security/*` bundles, independently publishable and loaded on demand. `install.ps1` / `install.sh` install everything and mount the preset and router automatically.
### Knowledge on demand Domain knowledge, rules, workflows and cases live in `references/` and are read on demand — never injected into the system prompt to decide for the model. Lean tokens, intact judgment.	### First-turn tool anchoring The first top-level request only exposes shell + `read`; the full catalog opens after promotion. A text-only first reply can't trap the session — request two always sees the full catalog.

## Why DSH security-analysis capability is scattered across domain bundles: `add` Android, `add` Web, `add` Native — and you still have to wire up the preset and router yourself. helmd packs six domains + evidence tooling + first-turn bootstrap into one preset: `one preset` `ten bundles` `zero manual wiring` Install once, send `helmd` in a session, and every domain's tools are ready. ## Architecture ```mermaid flowchart LR Q["Question"] --> SP["system-prompt/assemble"] SP --> BS["bootstrap first-turn narrowing
shell + read"] BS --> P{"promoted?"} P -->|no| BS P -->|yes| ROUTER["router domain routing"] ROUTER --> ANDROID["Android"] ROUTER --> WEB["Web"] ROUTER --> NATIVE["Native"] ROUTER --> PROTO["Protocol"] ROUTER --> MAL["Malware"] ROUTER --> AI["AI-Security"] ROUTER --> EVID["Evidence"] ANDROID & WEB & NATIVE & PROTO & MAL & AI & EVID --> REF["read_reference reads references/ on demand"] REF --> OUT["model's own judgment + confidence-scored conclusion"] style BS fill:#eff6ff,stroke:#3b82f6,stroke-width:2px,color:#1e40af style ROUTER fill:#f0fdf4,stroke:#16a34a,stroke-width:2px,color:#15803d style REF fill:#fffbeb,stroke:#f59e0b,stroke-width:2px,color:#92400e style OUT fill:#15803d,color:#fff,stroke:#166534,stroke-width:2px ``` - **First-turn narrowing**: the first top-level request only exposes shell + `read`; the full catalog opens after promotion - **Domain routing**: `router` routes problems via `skill_catalog` / `read_reference` - **On-demand references**: `references/` is a knowledge base, not an injection; the model reads and decides ## Quick start **Prerequisites**: a working [`dsh`](https://github.com/deepseek-ai/deepseek-harness) CLI and pnpm. Windows (double-click `install.bat`, or run via PowerShell): ```powershell .\install.ps1 ``` macOS / Linux: ```bash ./install.sh ``` The installer adds the 9 bundles, mounts the `helmd` preset, and sets the default in one shot. Then boot: ```bash dsh web ``` Send `helmd` in a session to activate. ## Verification ```bash dsh --profile web --dump-config # you should see the 9 @dsh-security bundle rows ``` After sending `helmd` in a session: ```text skill_catalog → returns the six-domain route map native_reference → reads the Native reference detect_packer → identifies the PE/ELF protector ``` Any of the above returning normally means the install succeeded. ## Packages | Package | Injects | Responsibility | Tools | | --- | --- | --- | --- | | `@dsh-security/bootstrap` | systemPrompt | First-turn tool-catalog narrowing, full catalog after promotion | none | | `@dsh-security/router` | tools | Domain routing and catalog | `skill_catalog`, `read_reference` | | `@dsh-security/skill-android` | tools | Android reversing | `apk_fingerprint` | | `@dsh-security/skill-web` | tools | Web security | `web_reference`, `bot_analyze` | | `@dsh-security/skill-native` | tools | Native / binary reversing | `native_reference`, `detect_packer`, `scan_strings`, `xor_bruteforce`, `encoding_detect` | | `@dsh-security/skill-protocol` | tools | Protocol / traffic | `protocol_reference`, `pcap_parse`, `state_machine`, `parse_har` | | `@dsh-security/skill-malware` | tools | Malware samples | `malware_reference`, `ioc_extract`, `yara_gen` | | `@dsh-security/skill-ai-security` | tools | AI / LLM security | `ai_reference`, `llm_sim` | | `@dsh-security/skill-evidence` | tools | Evidence / reporting / case | `evidence_reference`, `create_case`, `triage_artifact`, `hash_artifact` | | `@dsh-security/toolbox` | tools | Tool recommendation | `tool_recommend` | Each `*_reference` tool reads its own `references/` on demand; the entry point is its `index.md`. ## Alternatives | Approach | Why not | |----------|---------| | Install domain bundles by hand | 9 adds + manual preset + router wiring; repetitive and error-prone | | Raw shell tools only | no domain knowledge; the model guesses; unreproducible conclusions | | Stuff knowledge into the system prompt | token blow-up and it overrides the model's judgment | | helmd | one preset aggregates everything; knowledge read on demand | ## Common commands ```bash pnpm install # install deps (prepare auto-runs tsc) pnpm build # build all 9 bundles pnpm typecheck # tsc --noEmit type gate on a clean tree ``` Local tarball delivery: ```powershell .\scripts\repack.ps1 # emit dist-tgz\*.tgz dsh plugin --profile web add .\dist-tgz\*.tgz # install into the web profile ``` ## Deployment One-command install is above under Quick start; the manual steps are below. Prerequisites: the 10 `@dsh-security/*` packages published to npm (see Publishing). ### 1. Install the bundles into a profile ```bash dsh plugin --profile web add \ @dsh-security/bootstrap \ @dsh-security/router \ @dsh-security/skill-android \ @dsh-security/skill-web \ @dsh-security/skill-native \ @dsh-security/skill-protocol \ @dsh-security/skill-malware \ @dsh-security/skill-ai-security \ @dsh-security/skill-evidence @dsh-security/toolbox ``` `dsh plugin` forwards to pnpm inside the profile directory; the packages land in `$DSH_HOME/profiles/node_modules/`. ### 2. Mount the preset Copy `presets/full-reverse/` into the DSH user preset root `$DSH_HOME/.agent-presets/helmd/`: macOS / Linux: ```bash mkdir -p ~/.dsh/.agent-presets/helmd cp presets/full-reverse/agent.cordis.yml ~/.dsh/.agent-presets/helmd/ cp presets/full-reverse/preset.yml ~/.dsh/.agent-presets/helmd/ ``` Windows (PowerShell): ```powershell $p = Join-Path $env:USERPROFILE '.dsh\.agent-presets\helmd' New-Item -ItemType Directory -Force $p | Out-Null Copy-Item presets\full-reverse\agent.cordis.yml $p Copy-Item presets\full-reverse\preset.yml $p ``` ### 3. Set it as the default preset Pick `helmd` in the UI preset picker, or edit `$DSH_HOME/settings.yaml`: ```yaml agent-presets: default: helmd ``` ### 4. Boot and activate ```bash dsh web ``` Send `helmd` in a session to activate. `DSH_HOME` defaults to `~/.dsh`; substitute the path if you customized it. ## Directory layout ```text helmd/ ├── packages/ │ ├── bootstrap/ first-turn tool-narrowing filter │ ├── router/ domain routing and catalog │ ├── skill-ai-security/ AI / LLM security │ ├── skill-android/ Android reversing │ ├── skill-web/ Web security │ ├── skill-native/ Native / binary reversing │ ├── skill-protocol/ Protocol / traffic │ ├── skill-malware/ Malware samples │ ├── skill-evidence/ Evidence / reporting / case │ └── toolbox/ Tool recommendation ├── presets/ │ └── full-reverse/ └── docs/ ``` ## Build Requires pnpm; build targets ES2022 / NodeNext. ```bash pnpm install pnpm -r build ``` The root `pnpm build` is equivalent to `pnpm -r build`, running `tsc` per package; `pnpm typecheck` runs the `tsc --noEmit` type gate on a clean tree. ## Dependencies - `@deepseek-ai/cordis` `4.0.1` - `@deepseek-ai/dsh-tools` `0.1.0-rc.6` Versions are pinned via `overrides` in `pnpm-workspace.yaml`. ## Publishing - The root package is `private: true` and is not published; the 10 `@dsh-security/*` sub-packages are. - Each sub-package's `files` whitelist is limited to `dist`, `references`, `scripts`, `cordis.patch.yml`. - Each sub-package's `prepare` script runs `tsc` automatically before publishing. - Current version: `0.1.2`. ## Risks & mitigations | Risk | Mitigation | |------|------------| | DSH host upgrade breaks compat | peer deps on cordis / dsh-tools, pinned via `overrides` | | Missing `python` on the host | seam auto-probes python / py / python3, falls back to the `py` launcher | | Bundle/preset version drift | version 0.1.2, tarball and release published together | | Reference knowledge goes stale | read on demand, model's own judgment, non-binding | ## Acknowledgements This project integrates design ideas and implementation patterns from many excellent open-source projects, drawing on the experience of many community pioneers. Any resemblance is a tribute to great design. - [ADWMC/helm-x](https://github.com/ADWMC/helm-x) — prompt injection and scoring design - [yynxxxxx/Codex-X](https://github.com/yynxxxxx/Codex-X) — prompt templates and visual management ## Docs - [docs/principles.md](docs/principles.md) — design principles - [docs/architecture.md](docs/architecture.md) — architecture - [docs/architecture-v2.md](docs/architecture-v2.md) — architecture v2 (persona + tool anchoring + on-demand knowledge) ## Contributing Issues and PRs are welcome. Before making changes, read [docs/principles.md](docs/principles.md) and keep the architecture constraint: reference knowledge is read on demand and never overrides the model's own judgment. ## License Released under the [MIT License](LICENSE) — free to use, modify, and distribute. See [LICENSE](LICENSE). ## AI-generated & legal risk Some or all code in this repository was generated with AI assistance and may contain errors or be unsuitable for your context. Review it yourself and judge whether it fits your use case and jurisdiction before use; you are responsible for complying with local laws and for the consequences of using this project. Provided "AS IS" under MIT, without warranty of any kind.