Skip to content

AGENTS.md

Latest commit

 

History

History
89 lines (70 loc) · 2.63 KB

AGENTS.md

File metadata and controls

89 lines (70 loc) · 2.63 KB
Raw

AGENTS.md

Guidance for AI coding agents working with this repository.

Note: CLAUDE.md is a symlink to this file.

Repository Structure

This repo is a Claude Code plugin (d9-skills) and an Agent Skills repository.

.claude-plugin/
  plugin.json             # Plugin manifest
commands/
  skillops.md             # /d9-skills:skillops slash command
hooks/
  hooks.json              # SessionStart briefing + detect-xp reminder hooks
  session-briefing.sh
  detect-xp-reminder.sh
skills/
  {skill-name}/
    SKILL.md              # Operational skills (12)
  skillops/
    {skill-name}/
      SKILL.md            # Pipeline skills (8)
scripts/
  verify-skill-structure.mjs
  generate-skill-pr-comment.js

Skills Format

Skills follow the Agent Skills Open Standard.

Frontmatter (Required)

---
name: skill-name           # Required. Lowercase, hyphens only. Must match directory name.
description: ...           # Required. Max 1024 chars. What + when.
license: MIT               # Optional.
metadata:                  # Optional. Extra fields.
  version: "1.0.0"
  tags: [tag1, tag2]
  session_tokens: 5000     # Tokens consumed in the debugging session
  author: anonymous-contributor
  source: session-contribution
---

Body

Markdown instructions. Recommended sections:

  • Purpose — What problem this prevents
  • Triggers — When the skill should activate
  • Behavior — Step-by-step actions
  • Errors Prevented — Real error messages and descriptions
  • Restrictions — Hard/soft boundaries
  • Self-Check — Checkbox validation list
  • Examples — At least 2, including 1 edge case

Creating a New Skill

  1. Create directory: mkdir -p skills/{skill-name}
  2. Create SKILL.md following the format above
  3. Ensure name in frontmatter matches the directory name
  4. Open a PR

Skills are typically contributed via the SkillOps pipeline. Run /d9-skills:skillops (or /skillops from the main d9 repo) after a debugging session.

SkillOps Pipeline

The pipeline skills live in skills/skillops/ and handle the full contribution flow:

  1. detect-xp — Detect reusable patterns
  2. match-existing — Compare against existing skills
  3. anonymize-session — Strip PII and decontextualize
  4. format-skill — Generate SKILL.md
  5. refine-skill-design — Quality audit (ASQM >= 17)
  6. submit-skill — Clone, branch, push, PR
  7. curate-skills — CI scoring and audit
  8. discover-skills — Search external catalogs

Validation

node scripts/verify-skill-structure.mjs skills/{skill-name}
node scripts/verify-skill-structure.mjs --all