Claude Code Blueprint workflow
Claude Code Blueprint: 12 agents, 19 skills, 16 hooks and 6 CLAUDE.md rules, copy what you need or install a preset
- 73 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Hooks and agents install into `~/.claude/` and Claude Code settings, and hooks run shell scripts on every event
- The `setup.sh` script edits configuration on the machine and should be read before running
Install
- Works in
Manual install
mkdir -p ~/.claude && curl -o ~/.claude/CLAUDE.md https://raw.githubusercontent.com/faizkhairi/claude-code-blueprint/main/CLAUDE.mdFirst step: four behavior rules for all projects. Do not overwrite an existing CLAUDE.md without a backup.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Hooks and agents install into `~/.claude/` and Claude Code settings, and hooks run shell scripts on every event
- The `setup.sh` script edits configuration on the machine and should be read before running
This is third-party code. Review the repository files before installing.
What it does
The repository serves as a reference of Claude Code configuration: you copy files out of it into your own project and do not run Claude Code inside it. The first step takes a minute: `CLAUDE.md` goes into `~/.claude/` and sets four behavior rules, Verify-After-Complete, Diagnose-First, Plan-First and Verify-Before-Exit-Plan. The rest is added as the project grows: the `./setup.sh --preset=standard` installer applies the minimal, standard, core and full presets. The set includes 12 subagents with model tiering: project-architect on opus, the docs writer on haiku, the rest on sonnet. They include an architect, backend and frontend specialists, a code reviewer, a security reviewer, a DB analyst, devops, a tester and a memory curator. 19 skills trigger on plain phrases like "is this secure?" or "deploy": review, PR check, tests, planning, session saving, deployment. 16 hooks run on lifecycle events: blocking `git push`, secret scanning before commit, config protection, state saving before compaction and a final check on Stop. The author lists token costs: the base `CLAUDE.md` costs about 2,300 tokens per session, hooks use no context, and path-scoped rules load only when matching files are edited. Hooks are covered by 66 automated tests, and there is a beginner guide and a comparison with ten other tools.
Who it is for. Claude Code developers who want a proven configuration base, from a single file for a beginner to a full set of agents and hooks for a team.
Good fit when
- Claude Code says a task is done without checking the result and you need Verify-After-Complete and Diagnose-First rules
- You need to stop an accidental `git push` and commits with secrets using hooks that always fire
- A team of five assembles a shared set of agents and rules and commits it to the repository
- Context grows and you want to know how many tokens each configuration file costs
Not a fit when
- You need one ready plugin installed by command: the set is copied by hand or with the `setup.sh` script
- You want to run Claude Code inside the blueprint repository itself: it would load the blueprint's `CLAUDE.md` instead of yours
- The project runs wholly on another tool: some ideas carry over via the guide, but the files are built for Claude Code
Example request
Set up Claude Code Blueprint with the standard preset: CLAUDE.md in ~/.claude plus hooks for blocking git push and secret scanning, showing each stepLimitations
- The repository is a reference with no project template: you copy files into your own project and should not run Claude Code inside it.
- Every copied piece costs context: `CLAUDE.md` about 2,300 tokens per session, skills 480 to 1,070 tokens on demand.
- Agents are read-only by intent through `permissionMode: plan`, which holds only when the main session runs in default or dontAsk mode, and agents with memory can still write.
- Hooks and skills contain placeholders that you tailor to the project.
How to disable. Delete the copied files: `~/.claude/CLAUDE.md` and the hooks and agents you added from `~/.claude/` or the project's `.claude/`. The `setup.sh` installer changed only those directories. Remove hook entries from `settings.json` by hand.
FAQ
How do I install Claude Code Blueprint?
Download `CLAUDE.md` into `~/.claude/` with `curl -o ~/.claude/CLAUDE.md https://raw.githubusercontent.com/faizkhairi/claude-code-blueprint/main/CLAUDE.md`. The rest comes from `./setup.sh --preset=standard` in a clone. All variants are in `SETUP.md`.
What do the presets include?
There are four presets: minimal, standard, core and full. Contents are described in `docs/PRESETS.md`, and the full set includes 12 agents, 19 skills, 16 hooks and 6 rules.
How much context does it use?
The base `CLAUDE.md` costs about 2,300 tokens at session start. Hooks take no context, path-scoped rules load when matching files are edited, and a skill costs 480 to 1,070 tokens when its trigger phrase fires.
What are the four rules in CLAUDE.md?
Verify-After-Complete, Diagnose-First, Plan-First and Verify-Before-Exit-Plan. They make the agent check the result, diagnose a failure first, plan, and verify the plan before leaving plan mode.
Does the set work with other tools?
The files are built for Claude Code. `docs/CROSS-TOOL-GUIDE.md` covers Copilot, Cursor, Cline, Roo Code, OpenCode and five more tools: the ideas carry over, the files do not.
Discussion
No questions or reviews yet.
Related
WorkflowsA Claude Code working configuration: a planner with per-milestone review, sub-agents and skills for analysis, refactoring and docs sync
One-command Claude Code environment installer: a wizard sets up agents, rules, commands, skills and hooks in three profiles
A Claude Code environment installer: CLAUDE.md, rules, skills, agents, hooks and two-layer project memory in one script
A starter kit for Claude Code: a CLAUDE.md memory-bank template with ready hooks, subagents and skills you copy into your project
Shared Claude Code settings for a team: hooks against dangerous commands, a check before finishing and skills from spec to review
A handbook and starter set for Claude Code: guides, CLAUDE.md templates, starter kits, hooks and the commit-helper plugin.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Claude Code Blueprint (this card) | no | 73 | |
| My Claude Code Setup | no | 2,730 | one author's personal setup with no presets or hook tests |
| Claude Code Templates | no | 33,413 | a template catalog installing components, while here it is one vetted configuration |
README in short
The README explains that the repository is a reference of files to copy and warns against running Claude Code inside it. The quick start comes down to one command for `CLAUDE.md` and an installer with presets. Tables of token cost, who the set suits and how it differs from others follow. 12 agents with their models, 19 skills by category and 16 hooks by event are listed in detail.