Agent-Skills.md

Agent Skills: Skill: doc-update

Use when asked to update repository documentation, ADRs, or onboarding guidance.

UncategorizedID: Reneromero08/agent-governance-system/doc-update

Repository

Reneromero08/agent-governance-system
Reneromero08License: NOASSERTION

Install this agent skill to your local

pnpm dlx add-skill https://github.com/Reneromero08/agent-governance-system/tree/HEAD/CAPABILITY/SKILLS/utilities/doc-update

Skill Files

Browse the full folder contents for doc-update.

Download Skill

Loading file tree…

CAPABILITY/SKILLS/utilities/doc-update/SKILL.md

Skill Metadata

Name
doc-update
Description
"Use when asked to update repository documentation, ADRs, or onboarding guidance."
<!-- CONTENT_HASH: e55e25737af57b512b8918844cdd78f9ea09269c268c7be7975a15f31ba09547 -->

required_canon_version: >=3.0.0

Skill: doc-update

Version: 0.1.0

Status: Active

Trigger

Use when asked to update repository documentation, ADRs, or onboarding guidance.

Doc system overview (authoritative structure)

  • CANON/: binding rules (CONTRACT, INVARIANTS, VERSIONING). Only edit when explicitly asked.
  • CONTEXT/: ADRs, preferences, rejections, and research. ADRs explain why decisions exist.
  • CONTEXT/maps/: entrypoints and system maps that tell agents where to change things.
  • README.md: top-level overview and required session bootstrap.
  • MCP/: protocol integration docs and server entrypoints.
  • SKILLS/: procedural workflows + fixtures for non-trivial work.
  • CONTRACTS/: fixtures and schemas that enforce behavior.
  • TOOLS/: critics, linters, and migration helpers.

Inputs

  • input.json with:
    • topic (string): e.g. "mcp", "skills", "governance".
    • extra_targets (array, optional): Additional paths to include in the update set.

Outputs

  • Writes actual.json with:
    • topic (string)
    • doc_system (array of strings)
    • recommended_targets (array of strings)
    • notes (array of strings)

Workflow

  1. Confirm intent gate (do not edit CANON or existing CONTEXT unless explicitly requested).
  2. Use CORTEX/query.py to locate entrypoints; avoid raw filesystem discovery.
  3. Update the smallest set of docs that inform all agents (README, CONTEXT/maps/ENTRYPOINTS, MCP docs, relevant ADRs).
  4. For MCP updates, include:
    • The recommended entrypoint (LAW/CONTRACTS/ags_mcp_entrypoint.py).
    • The log location (LAW/CONTRACTS/_runs/mcp_logs/).
    • The verification skills (mcp-smoke, mcp-extension-verify).
    • Client config examples for Windows and WSL where relevant.
  5. Run python3 TOOLS/critic.py and python3 CONTRACTS/runner.py after changes.

Constraints

  • Must not scan the filesystem directly (use CORTEX/query.py for discovery).
  • Must not modify canon or context unless explicitly authorized.
  • Must not write artifacts outside allowed roots.

Fixtures

  • fixtures/basic/

required_canon_version: >=3.0.0