DeesseJS Skills

Canonical aggregation point for Claude Code skills authored by the deessejs GitHub organization.

Each skill lives in its own source repository under github.com/deessejs/<skill-name>. This repo mirrors them here so a single checkout gives an agent (human or AI) access to the full ecosystem.

Skills

Skill	Description
skills/create-issue/	Create GitHub issues with org-level type, Priority, Effort fields via REST API
[skills/research/]	Investigate a technical topic via web search: docs, forums, GitHub issues, best practices
[skills/diagnosing-bugs/]	Diagnose a GitHub issue: reproduce, find root cause, document findings, prepare for triage
skills/triage/	Triage a GitHub issue: analyze content, apply labels, post a structured review comment
skills/spec/	Explore a status:ready issue, write an implementation spec for human review, and push it to a branch
skills/implement/	Implement a spec-reviewed issue, or apply requested changes from a PR review
skills/create-pr/	Open a PR for an implemented issue — reads the spec, opens PR, updates labels, posts comment
skills/review-pr/	Review an open PR — check CI, validate against the spec, post a GitHub review
skills/release/	Complete a release — merge staging to main, generate changelog, clean up branches, post release notes
skills/ship/	Select which PRs go from staging to main, create release branch, open PR to trigger auto-release

Repository layout

.
├── README.md          # this file
├── AGENTS.md          # instructions for agents working in this repo
├── skills/            # one subdirectory per skill
│   └── <skill-name>/
│       └── SKILL.md   # the skill definition
└── .claude/           # local Claude Code config

For agents

When working in this repo:

1. Skill discovery — every folder under skills/ containing a SKILL.md is a loadable skill. List skills/ to enumerate what's available before assuming a task is not covered.
2. Skill format — each SKILL.md follows the Claude Code skill spec (YAML frontmatter with name, description; Markdown body with instructions).
3. Cross-skill work — when a task spans multiple skills, treat each skill's SKILL.md as the source of truth for its contract. Compose by invoking skills via the Skill tool.
4. Memory — persistent agent memory is at .claude/agent-memory/main/. Read it before making architectural assumptions.

Workflow

Bug reporté
     ↓
/diagnose #N      — diagnostiquer: reproduire, trouver la cause profonde
     ↓
/triage #N        — vérifier, dédupliquer, prioriser
     ↓
status:ready
     ↓
/spec #N          — planifier l'implémentation (approuvé par humain)
     ↓
/implement #N     — implémenter selon le spec
     ↓
/create-pr #N     — ouvrir la PR vers staging
     ↓
/review-pr #N     — revue + approbation
     ↓
Merge vers staging
     ↓
/ship             — sélectionner les PRs, créer release branch
     ↓
PR release→main  — merge déclenche auto-release
     ↓
/release          — finaliser: changelog, cleanup

Pour les questions techniques sans bug:

/research         — investiguer via web search
     ↓
Spec ou implémentation

Conventions

• Skill folder names are kebab-case and match the skill's name: in frontmatter.
• One skill = one folder. Do not bundle unrelated skills in the same folder.
• Keep SKILL.md self-contained: an agent reading only that one file should be able to use the skill.

Contributing

Skills are authored in their own repos under github.com/deessejs/<skill-name>. Edit upstream first, then re-sync here.

See AGENTS.md for full contributing guidelines.