The performance optimization system for AI agent harnesses. Skills, instincts, memory, security, and research-first development — battle-tested over 10 months of daily production use.
Up and running in under 2 minutes.
# Step 1: Install Plugin in Claude Code /plugin marketplace add https://github.com/affaan-m/everything-claude-code /plugin install everything-claude-code@everything-claude-code # Step 2: Clone & Install Rules git clone https://github.com/mrebrahim/everything-claude-code.git cd everything-claude-code ./install.sh --profile full # Step 3: Start Using /ecc:plan "Add user authentication"
.\install.ps1 --profile full
./install.sh typescript python
npx ecc-install typescript
Every component is designed to work together as a unified performance system.
183 skills covering TDD, security, frontend, backend, content, research, and deployment. Skills are scoped workflow bundles: reusable prompts, structure, and codemaps. The durable logic lives here, not in commands.
48 specialized agents that your orchestrator can delegate tasks to with limited scopes. Planner, architect, TDD guide, code reviewer, security reviewer, build resolver, and more. Each scoped with specific tool permissions.
Fire on PreToolUse, PostToolUse, Stop, SessionStart, and more. Auto-format after edits, warn on console.log, block git push for review, remind about tmux for long-running commands.
Modular .md files grouped by concern: security, coding-style, testing, git-workflow, agents, performance. Install only the languages you need: common + typescript/python/golang/swift/php.
1282 tests, 98% coverage, 102 static rules. Scans CLAUDE.md, MCP configs, hooks, and agents for secrets, injection risks, and vulnerabilities. Red-team/blue-team/auditor pipeline with Opus.
Instinct-based system automatically learns your patterns across sessions with confidence scoring. Import/export instincts, cluster them into skills with /evolve. Your setup gets smarter over time.
Quick reference for the most common tasks.
| I want to… | Command | Agent |
|---|---|---|
| Plan a new feature | /ecc:plan "Add auth" | planner |
| Design system architecture | /ecc:plan + architect | architect |
| Write code with tests first | /tdd | tdd-guide |
| Review code I just wrote | /code-review | code-reviewer |
| Fix a failing build | /build-fix | build-error-resolver |
| Run end-to-end tests | /e2e | e2e-runner |
| Find security vulnerabilities | /security-scan | security-reviewer |
| Remove dead code | /refactor-clean | refactor-cleaner |
| Review Python code | /python-review | python-reviewer |
| Review Go code | /go-review | go-reviewer |
From 10 months of daily Claude Code use.
Delete entire line — faster than backspace spam
Quick bash command prefix without leaving Claude
Search and reference files in your codebase
Multi-line input for complex prompts
Toggle thinking display on/off
Interrupt Claude and restore your code
Fork conversations for parallel non-overlapping tasks
Manually trigger context compaction at logical breakpoints
Monitor token spending during the current session
Between unrelated tasks — free instant context reset
// ~/.claude/settings.json { "model": "sonnet", "env": { "MAX_THINKING_TOKENS": "10000", // ~70% reduction in thinking cost "CLAUDE_AUTOCOMPACT_PCT_OVERRIDE": "50", // compact earlier = better quality "CLAUDE_CODE_SUBAGENT_MODEL": "haiku" // cheaper subagents } }
The same repo ships cleanly across all major AI coding harnesses.
| Feature | Claude Code | Cursor | Codex | OpenCode |
|---|---|---|---|---|
| Agents | 48 | Shared | Shared | 12 |
| Skills | 183 | Shared | 10 | 37 |
| Commands | 79 | Shared | Instruction-based | 31 |
| Hook Events | 8 types | 15 types | None yet | 11 types |
| MCP Servers | 14 | Shared | 7 | Full |