By pau-vega
A collection of developer workflow skills for planning, design review, and product requirements.
This skill should be used when the user asks to "grill me", "stress-test my plan", "challenge my design", "poke holes in this", "review my architecture", "play devil's advocate", "find the flaws", "what am I missing", or wants critical feedback on a plan, design, or idea before implementation.
This skill should be used when the user asks to "improve the architecture", "find refactoring opportunities", "deepen modules", "make the codebase more testable", "consolidate tightly-coupled modules", "review codebase architecture", or wants to surface architectural friction and propose module-deepening refactors.
This skill should be used when the user asks to "break down a PRD", "convert PRD to issues", "create issues from a PRD", "split a PRD into tickets", "create implementation tickets", or wants to turn a product requirements document into independently-grabbable GitHub issues.
This skill should be used when the user asks to "use TDD", "write tests first", "red-green-refactor", "test-driven development", "build this with tests", "fix this bug with TDD", or wants to implement features or fix bugs using a test-first approach.
This skill should be used when the user asks to "write a PRD", "create a product requirements document", "plan a new feature", "write requirements", "draft a spec", "spec out a feature", "define requirements", or wants to plan and define what to build before building it.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A curated collection of Claude Code plugins — typescript-rules, jsdoc-standards, and
workflow-toolkit — installable into Claude Code, Cursor, or OpenCode with a single
command.
The package is published to GitHub Packages, which requires authentication for both
public and private packages. Configure your ~/.npmrc once with a classic personal
access token scoped to read:packages:
@pau-vega:registry=https://npm.pkg.github.com
//npm.pkg.github.com/:_authToken=<your-classic-pat>
Then run the installer with npx:
npx @pau-vega/ai-devkit
Generate the token at GitHub -> Settings -> Developer settings -> Personal access tokens
(classic), with the read:packages scope only. Do not commit your ~/.npmrc to any
repo — this project's .gitignore excludes .npmrc for that reason.
If you would rather skip the token setup, npm can fetch the installer directly from this
repository over git:
npx github:pau-vega/ai-devkit
The trade-off: this clones the default branch HEAD (no version pinning), and the first run is slightly slower because npm has to resolve and fetch from GitHub directly. Functionally the installer behaves identically.
Ctrl-C at any prompt is a clean exit — no files are written until you confirm at the end of the prompt flow.
| Editor | project / project-local | user-global |
|---|---|---|
| Claude Code | <cwd>/.claude/ | ~/.claude/ |
| Cursor | <cwd>/.cursor/ | ~/.cursor/ |
| OpenCode | <cwd>/.opencode/ | ~/.config/opencode/ |
Inside each target the installer writes the standard subdirectories — skills/,
commands/, agents/, and either hooks/hooks.json (Claude Code, OpenCode) or
hooks.json at the root (Cursor). For project-local scope, the installer maintains a
delimited block in <cwd>/.gitignore listing every file it wrote, so re-running the
installer replaces the block in place and never duplicates entries:
# >>> AI-Devkit
.claude/skills/typescript-conventions/SKILL.md
.claude/commands/ts-review.md
...
# <<< AI-Devkit
hooks.json; it expects a
TypeScript plugin instead. Skills and commands are installed; real-time enforcement
hooks won't fire under OpenCode until a TypeScript port ships.agents/*.md from typescript-rules and jsdoc-standards is not copied
for Cursor targets. The skills and commands carry the relevant context.~/.cursor/hooks.json and
~/.cursor/hooks/ are formally documented. Commands, rules, and skills at user scope
follow the same shape but are not officially documented by Cursor.| Flag | Effect |
|---|---|
--dry-run | Print every file that would be written, without touching disk. Existing destinations are listed as would write (overwrites existing) — no prompts fire. |
--help | Show usage and exit. |
--version | Print the installer version and exit. |
Publishing happens automatically on a tagged GitHub Release. Cut a release in the GitHub
UI (Releases -> Draft a new release -> publish a vX.Y.Z tag). The
.github/workflows/publish.yml workflow runs npm pack --dry-run (sanity check) and
then npm publish against GitHub Packages, authenticated by the workflow's
GITHUB_TOKEN.
npx claudepluginhub pau-vega/ai-devkitStreamline git workflow with /commit, /commit-push-pr, and /clean-gone commands
JSDoc documentation standards with three enforcement levels (Minimal, Standard, Strict), automatic warnings, and on-demand review
TypeScript coding conventions with automatic validation, review command, and reviewer agent
PreToolUse hook to run TypeScript type-checks
TypeScript compiler validation and type checking for TypeScript projects.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review