Workflow plugins

Governance workflow skills: /session-start, /requirements, /design-first, /ux-design, /sprint, /automate, /uat, /session-end, /retro, /guide, plus the friction-capture memory loop.

10 skills

Working-protocol plugin: foreman orchestration, contrarian review, session-end, delivery-lifecycle, ADHD-shaped reporting, frontend design judgment, the three foreman agents, and two context-discipline hooks.

6 skills · 3 agents · hooks

House files (PLAN, TASKS, PROGRESS, DECISIONS) with verify, state, handoff and decision; plan-together; a fresh-context plan reviewer; and a frontend skill for UI that shows measured numbers.

9 skills · 1 agent

Innoveta's project-records methodology: tiered docs (CLAUDE.md + PROJECT_STATUS / ROADMAP / HISTORY), cold-start session handoffs, a parallel-safe ideas capture/triage workflow, and config-driven gated deploys.

10 skills

Spec-Driven Development pipeline: spec, plan, parallel implementation, review, tests and final verification

4 skills · 5 agents · hooks

Helpful agent skills.

10 skills

Rebuild a human's lost context on the work in flight after reconciling the current checkout's dangling changes — commit what belongs to an open PR, discard what already landed, sync the branch — then summarize the state of play, decisions, live PRs and tickets, and the single next step. Use when the user says "catch me up", "where were we", "what's the state of this", "get back on track", or invokes `/catch-up` after time away.

10 skills

Bring this session's branch up to date, land or discard its work, and archive the session.

10 skills

Dieci comandi per il lavoro su un repo (apertura del branch, convenzioni dei commit, squash della history, merge, release con changelog, triage dei rinvii, audit, verifica visiva della UI, diario di bordo, apertura di ticket Linear) e una card sopra il prompt con modello, contesto, limiti, branch, costo e tempo alla scadenza della cache. Le skill rilevano le convenzioni del repo invece di presumerle.

10 skills

Planning, code review, security review, TDD, build-error resolution, and cleanup agents plus stack-aware skills and commands for HalalWay projects.

9 skills

Route non-trivial tasks for maximum quality per token, with gates first, skeptics only where they pay, and a per-task ledger for follow-ups.

1 skill · 1 command · 6 agents · hooks

Local-SEO and accessibility skills for multi-location, server-rendered marketing sites: canonicals, host-derived indexing, JSON-LD, sitemaps, thin pages, WCAG 2.2 rules, plus a skill per page type and for the header and footer

9 skills

Model-gepinnte Subagents, selbstinstallierende git-Hooks, Secret/Prod-Guards, 3-Layer-Eskalation, Statusline-Tacho und Self-Update für Empire-Apps.

1 command · 7 agents · hooks

A three-stage engineering pipeline for Claude Code: /feature (security-aware design on your strongest model) → /feature-build (Opus-pinned plan + managed, reviewed execution with nested parallel tracks) → /feature-review (full security & reliability assessment). Includes skill-router nudges, [SEC]/[REL] flagging, build journals with resume, and Haiku-tiered verification.

5 skills · 3 commands · hooks

Skill akita-method + 8 slash commands /super-power:* (research, plan, start, phase, status, version, update, uninstall). Fluxo conversacional: coleta rica de contexto (docs/ + links sociais) → briefing extraído → pesquisa profunda customizada por tipo de projeto.

1 skill · 8 commands

Tool auto-approval with sensitive-path guards, Ollama LLM fallback, macOS notifications, an opt-in Stop-hook code review with pluggable reviewers (local LLM, generic CLI, in-session subagent), and a rule-adherence command that grades CLAUDE.md rules against past session transcripts

4 skills · 3 commands · 1 agent · hooks

Structured planning and task decomposition for complex, multi-step work: migrations, refactoring, multi-service features, with parallel execution and cross-session resume

2 skills · 3 commands · 4 agents

openspec/changes 全実装のオーケストレーション (単一セッション多段 SubAgent 方式 + Orca 並列方式)

4 skills · 5 agents

Rank every TODO item 0-5 by thinking required, then dispatch it to the cheapest model that can do it — five tier agents (haiku through opus@high), a /next dispatcher that reuses live agents, and the TODO.md marker grammar that carries the rank.

3 commands · 5 agents · hooks

Bootstrap a Fable-led working system into any session: seven pinned agents (scout, finder, builder, builder-tdd, verifier, adjudicator, synthesizer), an executable verify-claims workflow, refuter-verification doctrine, measured-numbers discipline, project scaffolding.

1 skill · 7 agents · hooks

Cycle SDD complet (commandes /sdd + /aiad), skills auto-déclenchées, agents de gouvernance Tier 1 (read-only, veto), hooks enforced (JNSP, Drift Lock, veto).

9 skills

Git-issue driven planning, orchestrator/worker development and verified merge-to-close for Claude Code (GitHub first)

6 skills · 3 agents

Portable agentic-coding extensions: slash commands, subagents, and skills for Claude Code, with /optimise auditing a project's Claude Code, Claude Design, or Codex setup.

3 skills · 6 agents

Enforces branch-from-dev workflow, conventional commits, secret scanning, docs & semver upkeep, and PR-only merges with human review

4 skills · 3 commands · hooks

Drive a GitHub milestone or issue set through small, independently mergeable slices with separated implementation, testing, and independent two-axis review.

3 skills · 4 agents · hooks

Teach repo agents to drive product-map: inspect what any repo ships, install pmap, keep manifests fresh, author local extractors, and run the Claude Design handoff loop.

8 skills

Injects portable engineering and communication preferences at session start, so they apply in every repo and in cloud sessions where ~/.claude is not yours.

6 skills · 1 agent · hooks

Plan-driven workflow: capture a plan in its own git repository, execute it with a delegated team in a branch or worktree of its own, and consume in-file markers.

7 skills · hooks

Finds, verifies, drafts, and opens PRs on approved open source repos, then babysits them for 30 days. Best-of-N drafting behind a reviewer gate, with a PR rate budget enforced at the tool boundary.

1 skill · 3 commands · 3 agents · hooks

KDC Club AI - AI-assisted coding skills for non-developers: project initialization, planning, implementation, and debugging with Python automation tools

4 skills · 4 commands