An agentic skills framework & software development methodology that works.
Portable skills for Claude Code and Codex
Blueprint packages ten focused skills that tell a coding agent what good work looks like across design, delivery, testing, review, and documentation. The skills are meant to be used only when needed, with repository policy kept in `AGENTS.md` and the mechanics left to the agent.
Builders who want reusable workflow skills for Claude Code, Codex, or other compatible coding agents.
You can guide your agent through clearer design, delivery, testing, and review steps without adding a heavy framework.
What it does
Architecture skill
Create or update `ARCHITECTURE.md` from verified code using `/architecture`.
Design skill
Write a technical design that is ready for review using `/design`.
Task to PR workflow
Turn decided work into a tested, independently reviewed pull request with `/task-to-pr`.
Planning support
Split decided work into smaller pieces with `/plan`.
Issue coordination
Coordinate a large batch of GitHub issues with `/codex-issue-coordinator`.
Check and improve
Use `/test`, `/review`, and `/improve` to validate and refine changes.
HTML doc output
Present work as HTML with `/html-doc`.
How to get it
- 1Install all ten skills with one command
npx skills add \ owainlewis/blueprint
README
Blueprint
Design. Plan. Build. Validate.
Why Blueprint · Install · Choose a skill · Read the guides · Contribute
Blueprint gives coding agents ten focused skills. They cover understanding existing code, deciding what to build, and delivering reviewed pull requests.
Why Blueprint
Capable agents do not need a script for every move. They need a clear outcome, the constraints that matter, and proof that the work is done.
Blueprint turns long-standing software engineering practice into a small set of skills: understand the system, make decisions before expensive changes, keep each task focused, test the result, and review the work. The skills say what good work looks like and what evidence is needed. They leave the mechanics to the agent.
The set is deliberately small. It contains the core skills its maintainer uses to make changes that another person can understand, test, and review. There is no agent framework to maintain. Install the skills and use only the ones the work needs.
Install
Install all ten skills with one command:
npx skills add \
owainlewis/blueprint
That is the main installation path. Blueprint works through portable skill files, so the same repository can support Codex, Claude Code, and other compatible coding agents without separate plugin workflows.
Choose a skill
Most work starts in one of these places:
- Document an implemented system. Use
/architectureto create or update rootARCHITECTURE.mdfrom verified code. - Decide how a meaningful change should work. Start with
/designfor a technical design ready for review. - Deliver a decided task. Start with
/task-to-prfor a tested, independently reviewed pull request.
Use /plan when decided work needs splitting. Use /codex-issue-coordinator when one Codex task must coordinate a large batch of GitHub issues.
Small, clear changes can go straight to /task-to-pr. Blueprint asks for only as much process as the work needs.
Ten focused skills
- Document:
/architecture - Decide:
/design,/architecture-review,/plan - Deliver:
/task-to-pr,/codex-issue-coordinator - Check and improve:
/test,/review,/improve - Present:
/html-doc
Each skill owns one engineering phase or useful outcome. Repository policy stays in AGENTS.md. Writing code, branching, debugging, and committing remain normal agent abilities inside the delivery workflow.
Guides
- Choosing the right skill explains where each skill starts and stops.
- Common Blueprint workflows shows how the skills fit together for small changes, larger features, existing systems, and issue batches.
- Examples contains reviewed designs and plans you can inspect or reuse.
- Migration guide explains how to remove older Blueprint skills before upgrading.
- Changelog records notable changes.
Project
Blueprint is deliberately small. It is not an issue tracker, an agent runtime, or a release system. It gives coding agents clear instructions for deciding, building, testing, and reviewing software.
Contributions are welcome. Read CONTRIBUTING.md before opening a pull request. Security problems should follow SECURITY.md.
Released under the MIT License.
Files in the repo
- .githooks
- .github
- assets
- examples
- guides
- scripts
- skills
- .gitignore
- AGENTS.md
- CHANGELOG.md
- CLAUDE.md
- CONTRIBUTING.md
- LICENSE
- MIGRATION.md
- README.md
- REVIEW.md
- SECURITY.md
Discussion (0)
Ask about usage, or say what you built with itSign in to join the discussion.
No comments yet. Be the first to say what this is good for.
More skills

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.
Topic in, narrated explainer video out. A Claude Code / Codex skill that turns any topic into a black-canvas motion-graphics explainer video with TTS voiceover, subtitles and a chapter progress bar. Chinese or English; every frame drawn in code with Remotion.
Public repository for Agent Skills
Open-source AI job search: scan job portals, evaluate listings into a structured A-H report with a global 1-5 score, tailor your CV, track applications — runs locally in your AI coding CLI (Claude Code, Codex, OpenCode, Antigravity…)

Production-grade engineering skills for AI coding agents.