Skip to content

About

A growing repertoire of self-authored Claude Code skills — spec, plan, build, PR-review, and debugging pipelines conducted by subagent ensembles — shipped as a single plugin.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

Repertoire

English | 简体中文 | 繁體中文 | 日本語

A collection of curated Claude Code skills, distributed as a single plugin. Install once to get all skills.

Install

/plugin marketplace add PythonShe/Repertoire
/plugin install repertoire@repertoire

Then restart or run /reload-plugins. Skills are namespaced under /repertoire:<skill>.

Recommended setup

These skills make extensive use of subagents and reasoning tokens. Recommended setup:

  • Claude Max (5x or 20x) or API billing: Recommended. Lighter tiers may reach rate limits quickly.
  • OpenAI codex-cc plugin (optional): Enables cross-model review for supported skills. Falls back cleanly to Claude-only review if not installed.
  • GitHub gh-stack extension (optional): Enables Maestro to link multi-plan sets into GitHub stacked PRs.
  • Model selection: Run sessions on /model best or opus (avoid Sonnet). Build tasks inherit your session model, while review tasks use Opus.

Skills

Skill Invoke What it does
Eureka /repertoire:eureka Explores and vets project ideas through structured brainstorming, producing a prioritized shortlist of viable concepts.
Libretto /repertoire:libretto Refines rough concepts into complete, implementation-ready technical specifications.
Score /repertoire:score Breaks down specifications into detailed implementation plans (including multi-plan sets for large projects).
Maestro /repertoire:maestro Executes implementation plans end-to-end, managing task implementation, code review, and quality verification.
Coda /repertoire:coda Triages and resolves open PR feedback, addressing review comments and fixing failing CI checks.
Encore /repertoire:encore Audits an existing codebase or module to discover and implement improvements in performance, security, and quality.
Presto /repertoire:presto Fast-track implementation for small, focused changes (≤150 lines) without requiring upfront specs or plans.
Jam /repertoire:jam Autonomous, unattended session that finds, implements, and verifies practical repository improvements.
Legato /repertoire:legato Audits and refines UI animations and transitions against professional motion standards.
Tuner /repertoire:tuner Diagnoses tricky bugs to their root cause and applies test-driven fixes.

Local development

Test changes directly from a clone:

claude --plugin-dir .
/reload-plugins

Repository layout

Repertoire/                       repo root = plugin root = marketplace root
├── .claude-plugin/
│   ├── plugin.json               plugin manifest
│   └── marketplace.json          marketplace catalog entry
├── skills/                       one directory per skill
│   └── <name>/                   eureka, libretto, score, maestro, coda, encore, tuner, presto, jam, legato
├── shared/                       shared subagent contracts and policies
├── docs/                         architecture decision records and authoring guides
├── CHANGELOG.md
└── README.md

Adding a new skill

  1. Create skills/<name>/SKILL.md (with name, description, and when_to_use in YAML frontmatter). Include trigger evals at skills/<name>/evals/evals.json.
  2. Add a row to the Skills table above.
  3. Validate and commit:
    claude plugin validate .

To release updates to existing installs, increment version in both .claude-plugin/plugin.json and .claude-plugin/marketplace.json.

About

A growing repertoire of self-authored Claude Code skills — spec, plan, build, PR-review, and debugging pipelines conducted by subagent ensembles — shipped as a single plugin.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors