Spec-driven Claude Code, dropped into your repo in one command.

---

Overview

For developers using Claude Code who want an opinionated starting point:

• A strict CLAUDE.md ruleset
• Five focused subagents — review, architect, simplify, TDD, security
• spec-kit wired up so non-trivial work goes through /specify → /plan → /tasks before any code is written

Works on macOS, Linux, and Windows (Git Bash or WSL).

---

Quick install

From inside the repo you want to set up:

curl -fsSL https://raw.githubusercontent.com/ifixai-ai/claude-strike/main/bootstrap.sh | bash

What it does

• Clones claude-strike into ~/.cache/claude-strike (or $CLAUDE_STRIKE_CACHE)
• Copies files into your repo
• Runs specify init --here --ai claude to add spec-kit

Re-running is safe. Existing files are backed up to <path>.bak.<timestamp> before being overwritten.

From a local checkout

git clone https://github.com/ifixai-ai/claude-strike.git /tmp/claude-strike
cd ~/your-project
bash /tmp/claude-strike/bootstrap.sh

Requirements

• A git repo — run git init first if needed
• uv on your PATH — see docs. If missing, files still install and spec-kit is skipped with a warning.

What gets written

Path	Purpose
CLAUDE.md	Ruleset Claude reads every session
.claude/agents/	Subagent definitions
.claude/rules/	Common and per-language coding rules
.specify/	spec-kit config and commands

---

Agents

Agent	Purpose
code-reviewer	Quality and security pass after writes
code-architect	Feature design and planning
code-simplifier	Cut complexity without changing behaviour
tdd-guide	Write tests first
security-reviewer	Deep security pass for auth, crypto, payments, user data

See CLAUDE.md for when to use each one.

---

Languages

Python has its own rules in .claude/rules/python/. Other languages use the common rules.

---

Security

See SECURITY.md to report a vulnerability.

License

MIT — see LICENSE. Upstream credits in NOTICE.