affaan-m/ECC — ECC – Agent Operating System for AI Coding
Transcript
QuickFacts
Welcome! Today we're exploring ECC, which stands for Everything Claude Code. It's an agent operating system that works across many different AI coding tools, providing hundreds of workflow skills and specialized agents to make your development sessions smarter and more organized.
PlainEnglish
Think of ECC as an operating system for AI coding assistants. Instead of manually configuring workflows for each tool, ECC gives you a shared catalog of skills like test-driven development, security review, and framework patterns. It supports fourteen different AI harnesses, from Claude Code to Cursor to Codex, and lets you install exactly the capabilities you need.
QuickFacts
ECC is built on a hybrid stack. The CLI and install layer runs on Node with TypeScript, handling manifest-driven installs and harness adapters. The ECC2 control plane is written in Rust, using ratatui for the terminal dashboard, tokio for async session management, and SQLite for tracking session state. It integrates with over thirty Model Context Protocol servers for things like GitHub operations, documentation search, and browser automation.
Architecture
Here's the overall architecture. The ecc CLI is your entry point for installing components, browsing the catalog, and launching the control pane. Install flows read manifests to discover which modules to copy, then harness adapters translate those modules into each tool's expected format. Skills and agents are the workflow content that gets installed. The ECC2 control pane, written in Rust, manages multiple sessions, tracks activity in a SQLite store, and provides a live dashboard.
Architecture
Let's trace an install flow. When you run ecc install typescript, the CLI consults the manifests to figure out which modules you need. It detects your harness by looking for specific config directories, like dot-claude or dot-cursor. The harness adapter then plans copy and merge operations, copying skills and agents, merging rule files, and updating hook configurations. Finally, it writes an install state file so it knows what's already installed.
PlainEnglish
Here's a helpful analogy. Skills are like recipe cards in a shared kitchen. Each skill is a Markdown file with frontmatter that says when to use it, what steps to follow, and what evidence to collect. Just like a recipe card tells you when to bake versus fry, a skill tells the AI assistant when to activate — for example, the TDD workflow skill activates when writing new features and enforces a strict test-first cycle with coverage checks.
Architecture
Agents work through delegation. The main AI session can call in specialist agents when it needs expertise. The planner agent breaks down complex features into implementation steps with dependencies. The code reviewer reads diffs and applies a strict checklist, only reporting issues it's confident about. The build fixer diagnoses errors and suggests solutions. Each agent does its work and reports back to the main session, which synthesizes the results.
Architecture
ECC2 is the control plane for managing multiple AI coding sessions at once. It's a Rust application with a terminal dashboard that shows you all your active sessions in real time. You can see which files each session is editing, how many tokens it's used, what the estimated cost is, and whether it's waiting for approval or stuck. It can create isolated git worktrees for each session, so every task runs in its own branch. There's even a daemon mode that watches for stalled sessions and auto-resumes them. It's still in alpha but already has nearly forty thousand lines of Rust.
PlainEnglish
Here's a typical journey with ECC. First, you install the CLI globally with npm. Then you add the components you need, like TypeScript or React skills. You can browse the catalog to discover what's available. When you start your AI coding session, skills auto-load based on your installed modules. As you work, you delegate complex tasks to specialist agents. And if you're juggling multiple sessions, the ECC2 dashboard keeps you on top of everything.
Architecture
ECC integrates with a wide ecosystem. It supports over thirty Model Context Protocol servers for things like GitHub operations, documentation search with Context7, browser automation with Playwright, and database access with Supabase. It works across fourteen different AI coding harnesses, from Claude Code to Cursor to Gemini. And it detects deployment platforms like Vercel and Netlify to provide workflow guidance. All of these integrations are managed through a central MCP config catalog.
PlainEnglish
That's ECC in a nutshell. One shared catalog of workflow skills and specialized agents that works across many different AI coding tools. Whether you're running Claude Code, Codex, Cursor, or something else, ECC gives you a unified way to install capabilities, delegate to experts, and orchestrate multiple sessions. It's like having an operating system for your AI assistants. Thanks for watching!
How this was made
Lenzon read affaan-m/ECC and generated this walkthrough automatically. The narration above is the transcript of what it says.
Explain a pull request from your own repo
Point Lenzon at a repo or a pull request and get a narrated walkthrough like this one.
Try it