affaan-m/ECC
A toolkit that teaches AI coding agents to work like disciplined engineers
ECC is an open-source system that makes AI coding tools like Claude Code, Codex, and Cursor follow a plan-test-build-review-verify-remember-improve workflow instead of freelancing on every prompt. It ships 68 agents, 286 skills, and 94 commands, and compresses each session's lessons into reusable summaries instead of storing raw transcripts. It's MIT-licensed and free, with a paid GitHub App available for private repos.
What it does
- Instead of retyping instructions like 'write tests first, review your work' in every prompt, you install that workflow once and it becomes standard behavior
- Code review happens from a fresh context separate from the one that wrote the code, so the same blind spots don't get missed twice
- Rather than saving entire chat transcripts, sessions get distilled into summaries and 'instincts' (patterns learned from real sessions with confidence scores) that carry over to future work
- A security scanner called AgentShield checks the agent's own configuration -- prompts, hooks, MCP settings, permissions, secrets -- as a potential attack surface
- Installs as a native plugin for Claude Code, with capability-limited adapters for Codex, Cursor, OpenCode, Gemini, and other harnesses
Why it matters
Developers who bounce between multiple AI coding tools and keep re-explaining the same rules can install this workflow once and share it across tools instead of repeating themselves. Treating an agent's own configuration files as a security surface is also a useful angle for teams rolling out AI coding tools together.
Terms in this repo
- Hooks · scripts that run automatically when specific events happen inside an AI coding tool
- Instincts · patterns learned from real work sessions, stored with a confidence score
- TDD (Test-Driven Development) · writing a failing test first, then writing code to make it pass
- AgentShield · a security tool that scans an AI agent's prompts, configuration, and permissions
- MCP · a standard protocol that lets AI models connect to external tools and data
Repository description (English)
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Open on GitHubTrending repos
- openai/codexOpenAI's coding agent, now runnable straight from your terminal
- cordiverse/cordisA TypeScript framework for snapping feature pieces on and off whenever you need them
- ripienaar/free-for-devA single repo lists every free-tier cloud, API and collaboration tool developers actually need
- Wei-Shaw/sub2apiA relay server that lets one Claude/OpenAI/Gemini/Grok subscription be shared by many users
- multica-ai/andrej-karpathy-skillsA single instruction file that stops AI coding assistants from making silent bad guesses
- eneskirca/nodetermA workspace that puts scattered terminal tabs and AI coding agents on one draggable map
- n8n-io/n8nn8n lets you build automation workflows and AI agents by dragging nodes on a canvas, with code as a fallback
- apache/makaAn AI agent workspace that keeps everything running on your own machine, with every action logged