Agent Plugins Marketplace
← All plugins

weed-harness

v6.0.0

Design artifacts plus shared progress-page delivery, model routing, completion gates, and Claude Code hooks. Design conversations and autonomous execution live in matt-loop.

Codex7 Skills

By weedmoLicense: MIT1 GitHub starsUpdated 3 days ago

Directory evidence

Runtimes
Codex
Parsed components
7 skill or MCP entries
Source updated
Sep 21, 2026
Manifest status
Canonical path parsed

The directory validates manifest shape and source location. It does not execute the plugin or provide a security endorsement. Review the indexing methodology →

Install weed-harness for Codex

Installs for the current user
codex plugin marketplace add weedmo/skills
codex plugin marketplace upgrade weed-plugins
codex plugin add weed-harness@weed-plugins

Paste and run these commands in a terminal with Codex. They add and refresh the weed-plugins catalog, then install this plugin.

Compatibility: the page URL and API slug “weed-harness” remain stable.

  • Codex: weed-harness@agent-plugin-marketplace → weed-harness@weed-plugins

The installer fetches third-party code from the source repository shown on this page. This directory validates manifest structure and source location, but does not perform a security audit; review the manifest, components, and source before installing.

Get the source manually
git clone https://github.com/weedmo/skills

Clone the source repository, then follow its setup instructions to add the plugin to a compatible client. The repository root is the plugin root.

Plugin files

weed-harness/
├── .codex-plugin/plugin.json
├── skills/autocode-board/SKILL.md
├── skills/design-map/SKILL.md
├── skills/interview-report/SKILL.md
├── skills/loop-gates/SKILL.md
├── skills/loop-report/SKILL.md
├── skills/model-routing/SKILL.md
└── skills/setup/SKILL.md

Included Skills7

autocode-boardskills/autocode-board/SKILL.md

The experiment board of the autocode loop — its view (`assets/view.html`), the data checks render.py runs before writing (`assets/validate.py`), and the templates, schemas, and verbatim prompts autocode reads at each step (`assets/reference.md`). Platform-neutral: both the Claude Code and the Codex editions of autocode render through this skill and loop-report. Reference skill: autocode names the section it needs; do not invoke it standalone.

design-mapskills/design-map/SKILL.md

Visualize the design discussed so far as an artifact page with architecture diagrams, decisions, alternatives, and open questions. Use when the user asks to see the current design as an artifact, diagram, or map. matt-design owns design interviews and spec confirmation; matt-auto owns implementation.

interview-reportskills/interview-report/SKILL.md

Renders matt-auto's decision log (question → decision → rationale, per pipeline stage) into the run's live decision-graph page: stages in order, each decision an editable node the user can rewrite or flag and export as <slug>.edits.json for matt-auto to rework from, plus ticket waves, review and PR lanes while the run executes. Owns the view and writes <slug>.data.json; build (render.py) and delivery (deliver.py) come from the shared `$loop-report` skill. Called by matt-auto after loading the confirmed design and before its final report — never standalone.

loop-gatesskills/loop-gates/SKILL.md

How the delegated loops (pr-babysit, autocode, matt-auto's ship step) use the unlazy skill so that "done" is a measured verdict: one runnable GATES.md ledger per unit, coordinator-approved, re-verified with gate-check.mjs, a retry bound, Orca boundaries. Reference skill, never invoked standalone.

loop-reportskills/loop-report/SKILL.md

Builds and delivers the live progress page of a long-running delegated loop (matt-auto, autocode, and any loop that runs for minutes to hours on the user's behalf) — one self-contained HTML the user can read while the loop runs and keep afterwards. The build is `assets/render.py` over `assets/shell.html` plus the calling loop's view; the delivery is `assets/deliver.py` (Orca artifact link, or the worktree's Orca browser tab when links are unavailable, or the path), which keeps the route stable for the whole run. Callers run those two scripts and relay the answer; they never run `orca` commands themselves. Called by interview-report (matt-auto's decision graph) and autocode (the experiment board); do not trigger standalone.

model-routingskills/model-routing/SKILL.md

Codex · OpenCode · Orca routing: cost/time-efficient pairs, Astra question answering, independent review, dispatch, and escalation. Never invoked standalone.

setupskills/setup/SKILL.md

Claude Code 전용 weed-harness 사용자 환경 셋업 — 터미널 UI(statusLine HUD)와 기본 설정(custom hooks 등록: language-rule, auto-update)만 담당. 스킬/플러그인 설치는 하지 않음. 멱등(idempotent)이라 여러 번 실행해도 안전. 트리거: '/setup', 'setup hud', 'setup hooks', 'statusLine 등록'.

Plugin manifests1

.codex-plugin/plugin.json
{
  "name": "weed-harness",
  "version": "6.0.0",
  "description": "Design artifacts plus shared progress-page delivery, model routing, completion gates, and Claude Code hooks. Design conversations and autonomous execution live in matt-loop.",
  "author": {
    "name": "weedmo",
    "url": "https://github.com/weedmo"
  },
  "homepage": "https://github.com/weedmo/skills",
  "repository": "https://github.com/weedmo/skills",
  "license": "MIT",
  "keywords": [
    "codex",
    "skills",
    "harness",
    "loop"
  ],
  "skills": "./skills/",
  "interface": {
    "displayName": "weed-harness",
    "shortDescription": "Visual design plus shared progress, routing, and completion runtime",
    "longDescription": "Visualize existing designs with design-map. Shared runtime for matt-design/matt-auto and auto-loop: progress pages, model routing, and completion gates. design-map no longer interviews, confirms specs, or implements.",
    "developerName": "weedmo",
    "category": "Coding",
    "capabilities": [
      "Read",
      "Write"
    ],
    "brandColor": "#2563EB",
    "screenshots": []
  }
}

For maintainers

If you maintain this plugin, link to this source-backed listing from your README so users can review its manifest and indexed components.

[weed-harness on Agent Plugins Marketplace](https://pluginsmp.com/plugins/weed-harness)