AI analysis grounded in the code graph — computed facts, not vibes · 2026-09-08T03:03:19Z
ECC ("Everything Claude Code") is a distribution of configuration, tooling and automation for AI coding agents — skills, hooks, MCP manifests, memory-persistence logic and security scanners — designed to be installed across Claude Code, Codex, OpenCode, Cursor and similar harnesses. It ships as an installable package (npm packages ecc-universal, ecc-agentshield) plus a large scripts/ directory that handles install-target detection, session/state management and a "skill-evolution" versioning system, rather than being a single application.
The growth aligns with a real release cadence: v1.9.0 (selective install, 12 language ecosystems), v1.10.0 (operator workflows, ECC 2.0 alpha), v2.0.0 (control-pane, 261 skills, worktree lifecycle service) and v2.1.0 (Plan Canvas, Kimi Harness, self-hosted compute) all landed within roughly five months, each adding user-visible capability rather than just patches. The README's claim of "140K+ stars, 21K+ forks, 170+ contributors" and Anthropic hackathon origin, combined with active bug-fixing across multiple harness integrations (OpenCode crash fixes, Claude skill install flattening, cost-tracker inflation fix), suggests the growth is tied to concrete cross-tool compatibility work rather than marketing alone — though the star count itself (234,908, +636/day) is large enough that some of it likely reflects viral/social sharing (e.g. X/Twitter guide threads) not fully evidenced in the graph facts.
What changed recently, how it's actually built (from the code graph), and whether you should care. Free account — no card, no spam.