Ship Shit Dev Skills

<!-- catalog-summary:start -->

164 AI agent skills for development workflows. Works with Claude Code, OpenAI Codex, and Cursor.

Catalog: 164 skills · 30 commands · 13 bundles · 177 plugins.

<!-- catalog-summary:end -->

Skills are model-agnostic playbooks: the harness supplies the model, so no skill names a concrete model — orchestrators speak in capability tiers, and each repo's routing block maps tiers to models. Enforced by scripts/validate-skill-sync.sh; standards live in .agents/memory/system/skill-standards.md.

Directory Structure

<!-- catalog-layout:start -->

Path	Role	Generated fact
.agents/	Repository memory, standards, and maintenance skills	Tracked
.claude/	Claude loader adapters for shared maintenance content	Tracked
.claude-plugin/	Generated Claude marketplace catalog	177 generated plugins
.codex/	Codex loader adapters for shared maintenance content	Tracked
.github/	Issue templates and GitHub Actions workflows	Tracked
.husky/	Git hook configuration	Tracked
assets/	Static repository assets	Tracked
bundles/	Generated marketplace bundle snapshots	13 generated bundles
commands/	Claude Code/plugin command adapters	30 command adapters
prompts/	Shared prompt resources	Tracked
resources/	Authoring references and supporting documentation	Tracked
scripts/	Validation, generation, migration, and audit tooling	Tracked
skills/	Canonical public Agent Skills sources	164 canonical skills

<!-- catalog-layout:end -->

What's Included

• The Dev Loop: a board-driven autonomous workflow that ships software with gh + Claude + Codex.
• Engineering Workflows: code review, debugging, QA, release, GitHub triage, branch cleanup, production audit, and CI repair.
• Implementation Specialists: frontend, backend, infrastructure, security, testing, payments, and workspace setup skills.
• Agent Engineering: context engineering, memory systems, evaluation, multi-agent patterns, MCP building, and prompt/tool design.
• Maintenance Tooling: scripts that validate skills, regenerate bundles, and keep the marketplace catalog aligned with source.

The Dev Loop — ship software with gh + Claude + Codex

The flagship workflow: a board-driven autonomous dev loop that turns a GitHub issue into a reviewed PR, with you as architect/reviewer. It is the open, gh-driven version of ShipCode's pipeline — same stages, no app required.

• Plan (human): feature-intake → prd-writer / writing-plans write the PRD and implementation plan onto the issue.
• Dispatch (human opt-in): apply dispatch:claude (Claude lane), dispatch:codex (Codex lane), or dispatch:openrouter (OpenRouter lane) to a Backlog issue.
• Execute (AI): the loop claims it → In Progress → branch → implement → qa-reviewer + tests → PR, advancing loop:* phase labels as it goes.
• Review (human): the PR lands in Human Review, auto-assigned to you. Merge = Done; reject = back to Backlog.

Board columns are for humans (Backlog · In Progress · Human Review · Done · Deferred); the AI loop's sub-phases ride as loop:* labels. Status is the GitHub Projects board Status field — the single source of truth.

# 1. Install the dev-loop bundle (every skill the loop needs)
/plugin marketplace add shipshitdev/skills
/plugin install shipshitdev-dev-loop@shipshitdev

# 2. Provision your repo: labels + board + workflows + secrets (idempotent)
bash scripts/setup-dev-loop.sh            # --dry-run to preview

# 3. Write your repo's routing block (tracker + labels + domain). In Claude Code:
/setup-agent-routing

# 4. Drive it
/loop                                     # Phase 1: local pull (Claude lane)
# …or apply dispatch:claude / dispatch:codex to fire Phase 2 (GitHub Actions)

Full reference: .agents/memory/system/ai-dev-loop.md.

Installation

Quick Install (Recommended)

# Install all skills globally for Claude Code, Codex, and Cursor
npx skills add shipshitdev/skills -g --agent claude-code codex cursor --skill '*' -y

# Install specific skills
npx skills add shipshitdev/skills -g --skill stripe-implementer -y

# List available skills
npx skills add shipshitdev/skills --list

Do NOT use --all — it installs to every agent the CLI knows about (30+). Always use --agent to target only the agents you use.

Migrating from the retired shell installer

The legacy scripts/install-skills.sh installer was removed. It maintained a second hardcoded bundle inventory, targeted Claude-only project paths, and could recursively replace real directories during install or restore. Do not restore or run an old copy.

Migrate without deleting existing directories first:

1. Run the supported Quick Install command above with an explicit --agent list. The skills CLI validates requested names against the repository catalog and owns the managed installation layout.
2. Confirm the expected skills appear in each selected agent. Keep the previous directories in place until that verification is complete.
3. Review old links/directories individually. Remove only a path you have identified as an obsolete managed link; never recursively delete an agent skills directory as part of migration.

The README installation commands and catalog.json are the installation/catalog source of truth. Marketplace bundles remain available through the Claude Code plugin alternative below.

Project-local Install

npx skills add shipshitdev/skills --agent claude-code codex cursor

Claude Code Plugin (Alternative)

/plugin marketplace add shipshitdev/skills
/plugin install shipshitdev-dev-loop@shipshitdev
/plugin install shipshitdev-dev-workflow@shipshitdev
/plugin install shipshitdev-github@shipshitdev
/plugin install shipshitdev-testing@shipshitdev
/plugin install shipshitdev-security@shipshitdev

Platform setup boundary

The portable workflow source is always skills/<name>/SKILL.md. Platform entry surfaces should route to that source instead of copying it.

• Codex project instructions: use AGENTS.md and optional nested AGENTS.override.md files. Global preferences live in ~/.codex/AGENTS.md.
• Codex configuration: user defaults live in ~/.codex/config.toml; trusted repositories may add .codex/config.toml overrides. Model, effort, approvals, sandboxing, workspace, and other execution settings belong there or in the app.
• Codex workflows: install shared skills under .agents/skills. Do not add .codex/instructions.md or .codex/commands; neither is a supported project instruction or reusable workflow surface.
• Deprecated Codex prompts: ~/.codex/prompts is user-local, explicit-only, and deprecated in favor of skills. This repository does not publish prompts there.
• Claude commands: files under commands/ are thin Claude Code/plugin conveniences. They are not included in generated bundle manifests and are not presented as Codex commands unless a supported adapter is added later.

For Contributors

git clone https://github.com/shipshitdev/skills.git ~/shipshitdev-skills
cd ~/shipshitdev-skills
npx skills add . -g --agent claude-code codex cursor --skill '*' -y

Adding Skills & Commands

Adding a Skill

1. Create directory in skills/skill-name/
2. Add SKILL.md with YAML frontmatter
3. Update this README

mkdir -p skills/my-skill
touch skills/my-skill/SKILL.md

Adding a Command

Commands are Claude Code/plugin entry points, not portable workflow definitions. Route each command to a canonical skill whenever the workflow should work in Codex or another Agent Skills-compatible harness.

1. Create a thin .md file in commands/
2. Follow naming: {verb}-{noun}.md
3. Update this README

Documentation

• .agents/memory/system/skill-standards.md - Agent Skills spec + Claude Code extensions
• .agents/memory/system/skill-management.md - Single-source skill workflow
• .agents/memory/system/architecture.md - .agents folder structure
• .agents/memory/system/platform-adaptations.md - Claude vs Codex writing guide
• .agents/memory/system/execution-boundary.md - app-owned execution settings vs reusable content
• .agents/memory/system/routine-standards.md - portable scheduled-task and automation authoring
• .agents/memory/system/ai-dev-loop.md - Board-driven autonomous dev loop (/loop + agent-dispatch)

Development Commands

Command	Description
address	Resolve PR review comments — propose fixes and replies
agent	Audit, configure, scaffold, and route agents and subagents
board	Set up, audit, or normalize a GitHub Projects v2 board
bug	File a GitHub bug issue
clean	Clean completed tasks and consolidate session files
codex-loop	Claim and work one dispatch:codex issue locally via codex exec
deploy	Deploy the app and provision infrastructure
design	Review and refine UI — audit, critique, polish, layout
deslop	Remove AI slop and tells from code
env	Scaffold and validate environment variables
feature	Capture requirements into PRD epics and GitHub issues
fix-ci	Diagnose and fix failing CI checks on a PR
inbox	Capture quick tasks into a project inbox
loop	Claim and work one dispatch issue end-to-end
merge	Review and land open PRs into the trunk
performance	Analyze frontend, backend, database, and infrastructure performance
pr	Create, update, and publish a pull request
prd	Create specs, PRDs, and feature plans
prompt	Optimize prompts for AI generation
qa	Run a structured verification pass before commit
release	Cut a trunk release with patch notes
review	Review changes, a PR, all open PRs, or recent commits
scan	Run dependency, code, config, and OWASP security scans
skill	Author, capture, test, and scout agent skills
standup	Summarize what you shipped over a time window
suggest	Post inline suggested changes on a PR
test	Run, author, and set up tests
tests	Run the right tests and turn red green

<!-- catalog-skills-heading:start -->

Skills (164)

<!-- catalog-skills-heading:end -->

Dev Loop (10)

interview, feature-intake, prd-writer, writing-plans, prd-quality-gate, prd-task-creator, executing-plans, setup-agent-routing, gh-project-board, qa-reviewer

Dev Workflow (41)

agent-architecture-audit, agent-config-audit, ai-agent-cost-optimizer, ai-regression-testing, analyze-codebase, codebase-advisor, code-review, structural-review, full-code-review, review-dispatch, commit-summary, changelog-generator, standup, de-slop, refactor-dispatch, tech-debt, stack-modernization, debug, deploy, execution-debugging, deployment-composer, docs, llm-structured-output, merge-open-prs, production-audit, refactor-code, release, release-dispatch, release-cleanup, release-pr-gates, scaffold, shape, skill-capture, skill-comply, skill-scout, systematic-debugging, receiving-code-review, verification-before-completion, worktree, skill-dispatch, deploy-dispatch

GitHub (19)

bug, gh-address-comments, gh-fix-ci, gh-inbox, gh-pr-publish, gh-project-board, gh-review-suggestions, github-actions-author, git-safety, feature-intake, merge-open-prs, release, release-dispatch, release-cleanup, release-pr-gates, worktree, finishing-a-development-branch, pr-comments, fix-merge-conflicts

Testing (8)

playwright-e2e-init, tdd, testing-expert, testing-cicd-init, qa-reviewer, husky-test-coverage, test-runner, test-dispatch

Frontend & React (32)

frontend-design, component-library, accessibility, audit, clarify, critique, design-consistency-auditor, deslop-ui, html-style, layout, polish, quieter, react-component-performance, react-hook-form, theme-factory, react-patterns, react-refactor, react-testing-library, react-native-components, expo-architect, landing-page-vercel, micro-landing-builder, ai-loading-ux, table-filters, quick-view, nextjs-validator, nextra-writer, shadcn, shadcn-setup, tailwind, tailwind-validator, design-dispatch

Backend & Data (8)

api-design-expert, error-handling-expert, graphql-architect, nestjs-expert, incremental-fetch, turborepo, typescript-expert, typescript-refactor

Infrastructure (12)

docker-expert, aws-infrastructure, ec2-backend-deployer, vercel-deploy, postgres-ops, mongodb-migration-expert, mongodb-atlas-checker, monitoring-setup, nestjs-queue-architect, performance-expert, redis-caching, security-expert

Security (5)

security-audit, security-expert, git-safety, dependency-audit, open-source-checker

AI Agents (17)

prompt-engineering, mcp-builder, skill-creator, context-fundamentals, context-optimization, context-degradation, context-engineering, memory-systems, multi-agent-patterns, tool-design, evaluation, advanced-evaluation, comment-mode, spec-first, agent-browser, agent-dispatch, codex-image-gen

Workspace Setup (11)

fullstack-workspace-init, project-init-orchestrator, linter-formatter-init, clerk-validator, content-script-developer, package-architect, artifacts-builder, open-source-checker, devcontainer-setup, biome-validator, bun-validator

Planning & PRDs (12)

icp, roadmap-analyzer, roadmap-to-milestones, cto-advisor, interview, feature-intake, prd-task-creator, prd-writer, prd-quality-gate, rules-capture, writing-plans, prd-dispatch

Payments & Product Integrations (1)

stripe-implementer

Session Management (7)

agent-folder-init, session-end, session-start, session-documenter, executing-plans, setup-agent-routing, workspace-performance-audit

How Skills Adapt to Projects

Skills are adaptive - they scan project documentation to understand:

• Project architecture and structure
• Tooling, scripts, and CI conventions
• Existing patterns and conventions
• Domain terminology and implementation constraints

If a project has its own skill, the generic skill will collaborate with or defer to it.

Publishing & CI/CD

When you push to master, GitHub Actions automatically regenerates the bundles/ directory to keep marketplace plugins in sync with skills.

Claude Marketplace

Users install directly from GitHub:

# Add the marketplace
/plugin marketplace add shipshitdev/skills

# Install category bundles
/plugin install shipshitdev-dev-loop@shipshitdev
/plugin install shipshitdev-dev-workflow@shipshitdev
/plugin install shipshitdev-github@shipshitdev
/plugin install shipshitdev-testing@shipshitdev
/plugin install shipshitdev-frontend@shipshitdev
/plugin install shipshitdev-backend@shipshitdev
/plugin install shipshitdev-infrastructure@shipshitdev
/plugin install shipshitdev-security@shipshitdev
/plugin install shipshitdev-session@shipshitdev
/plugin install shipshitdev-planning@shipshitdev