architecture · owainlewis — Claude Code skill: what it does, how to install
A skill for Claude Code and Codex that writes or updates the root ARCHITECTURE.md file from a fixed template. It locks in system boundaries, data flow diagrams, and architectural rules.
MIT 331 installs ★ 438 updated 2026-11-10
npx skills add owainlewis/blueprint --skill architecture -a claude-codeWhat architecture does
The skill inspects REQUIREMENTS.md, existing code, and infrastructure schemas, then interviews you about system boundaries and data ownership. Once clarified, it fills out ARCHITECTURE.md with tradeoffs, rejected alternatives, and invariants.
Fabricating workload or latency figures is strictly forbidden: unknown targets stay recorded as open decisions. The skill also halts immediately after preparing the document for human review, without drafting implementation tasks or writing code.
What it helps with
- draft a root ARCHITECTURE.md document from project requirements
- document system boundaries and data ownership across services
- map data flow diagrams and entity models without manual formatting
- track architectural decisions alongside rejected alternative options
Example prompts for the agent
- Generate ARCHITECTURE.md using REQUIREMENTS.md and the current codebase
- Update our architecture document to reflect the new authentication service
- Draft the data model and flow diagram for this repository
How to install architecture
1Skills CLI (Vercel)
npx skills add owainlewis/blueprint --skill architecture -a claude-code2Manually: put the files in place
~/.claude/skills/architecture/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/architecture/
1Skills CLI (Vercel)
npx skills add owainlewis/blueprint --skill architecture -a codex2Manually: put the files in place
.agents/skills/architecture/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add owainlewis/blueprint --skill architecture -a cursor2Manually: put the files in place
.cursor/skills/architecture/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install owainlewis/blueprint skills/architecture2Manually: put the files in place
.github/skills/architecture/SKILL.md1Skills CLI (Vercel)
npx skills add owainlewis/blueprint --skill architecture -a gemini-cli2Platform CLI
gemini skills install https://github.com/owainlewis/blueprint --path skills/architectureManually: put the files in place
.devin/skills/architecture/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add owainlewis/blueprint --skill architecture -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about architecture
How do I install architecture in Claude Code?
Run npx skills add owainlewis/blueprint --skill architecture -a claude-code.
What does the architecture skill do?
It structures your root ARCHITECTURE.md file using an architecture template, detailing system decisions, invariants, and data diagrams.
Does the skill write implementation code after designing the system?
The instructions explicitly prevent code writing and task breakdown, stopping work as soon as the architecture document is ready for human review.
Source text
SKILL.md · MIT
--- name: architecture description: "Writes root ARCHITECTURE.md from a fixed template: the intended system, a data flow diagram, the data model, and the rules every feature must follow. Use for a new system or a change to its boundaries, data, or deployment." user-invocable: true argument-hint: "<system, requirements, or ARCHITECTURE.md>" disable-model-invocation: true --- # Architecture Fill in the [architecture template](references/architecture.md) in root `ARCHITECTURE.md`. It describes the system you intend to build and stays current as it changes. Product needs go in [requirements](../requirements/SKILL.md); one feature's design goes in a [spec](../spec/SKILL.md). ## Steps 1. Read `REQUIREMENTS.md`, the repository instructions, and any existing code, schemas, and infrastructure. Check every claim about the current system against them. 2. Ask the user about each decision that shapes the system: boundaries, who owns which data, dependencies, and hard limits. Give your recommended answer with each question. 3. Choose the simplest design that meets the requirements. For each big decision, name the strongest alternative and why you did not pick it. 4. Fill in the template. Replace every bracketed placeholder and delete the guidance comments. Render the diagram and check that it matches the text. 5. Stop when the document is ready for review. Do not plan tasks or write code. ## Template Use [references/architecture.md](references/architecture.md). ## Rules - Keep existing `INV-n` IDs. Never reuse or renumber them. - Set the status to Approved only when a human approves the document. Approved is not the same as built. - Take the author from the request, or else from `git config user.name`. Set the date each time the document changes. - Never invent workload or latency numbers. Mark unknown targets as open decisions. - Update this document when a change moves ownership, data, interfaces, trust boundaries, or deployment. ## Return Report the path, the main decisions and their tradeoffs, and any blocking question.
© owainlewis · MIT · owainlewis/blueprint
Similar skills
All skills for Writing and docs →architecture by other authors comes first.