ECC is the harness-native operator system for agentic work — built from 10+ months of intensive daily use building real products across multiple AI coding environments. It is not a config pack. It is a complete system: skills, hooks, agents, memory optimization, continuous learning, security scanning, and multi-agent orchestration workflows that survive across context resets and harness upgrades. The entire stack — 261+ skills, 67 agents, 93 legacy command shims, and theDocumentation Index
Fetch the complete documentation index at: https://mintlify.com/affaan-m/ECC/llms.txt
Use this file to discover all available pages before exploring further.
orch-* orchestrator family — installs into your existing harness without replacing it.
Official sources only. Install ECC exclusively from verified channels: the GitHub repository github.com/affaan-m/ECC, the npm package
ecc-universal, the ecc-agentshield package, the GitHub App, the plugin slug ecc@ecc, and the project website ecc.tools. Third-party re-uploads and unofficial mirrors are not maintained or reviewed by the project and may contain malware.What ECC Is
Most AI coding harnesses ship with a generic context window and no built-in workflow structure. Developers end up reinventing the same patterns — test discipline, security checklists, build-error recovery, parallel agent delegation — on every new project and for every new tool they adopt. ECC solves this once, harness-wide. The system comprises five core primitives that layer on top of whatever harness you already use:Skills
The primary workflow surface. Skills are scoped, reusable workflow bundles — structured prompts, supporting files, and codemaps — that the agent can invoke, suggest, or compose autonomously. Examples include
tdd-workflow, security-review, verification-loop, and deep-research. Skills replace ad-hoc prompting with reproducible, version-controlled workflows.Hooks
Trigger-based automations that fire on harness tool events. Hooks run before or after tool calls, on session start and stop, on context compaction, and on user prompt submission. They enforce code quality, persist session memory, gate destructive commands, and detect secrets — automatically, without manual invocation.
Commands
Maintained slash-entry compatibility for command-first workflows. Commands like
/plan, /code-review, /build-fix, and /quality-gate wrap the underlying skill logic and remain available during migration. New workflow development lands in skills first; commands serve as durable entry points while the surface evolves.Agents
67 specialized subagents that the orchestrator can delegate bounded tasks to. Each agent —
planner, architect, tdd-guide, code-reviewer, security-reviewer, build-error-resolver, and more — runs with scoped tools, limited permissions, and focused instructions. Agents parallelize work and keep the main context clean.Memory
Session lifecycle hooks persist context across resets. The continuous-learning system extracts patterns from sessions into reusable instincts with confidence scoring. A SQLite state store tracks install state, session history, and skill-run health. Memory keeps ECC’s knowledge of your codebase alive between sessions.
Rules
Always-follow guidelines organized into
common/ (language-agnostic) and language-specific directories — TypeScript, Python, Go, Swift, PHP, ArkTS, and more. Rules encode security requirements, testing discipline, commit format, and design patterns so Claude applies them to every response without being asked.Supported Harnesses
ECC installs natively across 14 harness targets, each with a dedicated adapter. The same skill library, agent definitions, and workflow patterns work regardless of which AI coding tool you use.| Target | Description |
|---|---|
claude | Claude Code CLI — the primary native target |
claude-project | Claude.ai project-level install |
cursor | Cursor IDE with 15-event hook adapter (DRY pattern) |
codex | OpenAI Codex app and CLI via AGENTS.md and .codex/ |
gemini | Gemini CLI via .gemini/GEMINI.md |
opencode | OpenCode full plugin via .opencode/ with 20+ hook events |
antigravity | Antigravity IDE with .agent/ adapter |
zed | Zed editor via .zed/ project adapter |
kimi | Kimi via .kimi/ adapter |
qwen | Qwen CLI with home-directory selective install |
codebuddy | CodeBuddy project-local selective install |
hermes | Hermes operator workflow surface |
openclaw | OpenClaw via .openclaw/ adapter |
joycode | JoyCode project-local adapter |
Version History
v2.0.0 — The Agent Harness Operating System (Jun 2026)
Stable graduation of the 2.0 line: 261 skills, the control-pane substrate (session adapters + MCP inventory), the worktree-lifecycle service, theorch-* orchestrator family, and the launch of the ECC Discord community. This is the version described throughout these docs.
v1.9.0 — Selective Install & Language Expansion (Mar 2026)
Manifest-driven install pipeline withinstall-plan.js and install-apply.js for targeted component installation. SQLite state store for install tracking and incremental updates. Six new language-specific agents (TypeScript, PyTorch, Java, Kotlin). Twelve language ecosystems including Java, PHP, Perl, Kotlin, C++, and Rust.
v1.8.0 — Harness Performance System (Mar 2026)
ECC formally framed as an agent harness performance system. Hook reliability overhaul: SessionStart root fallback, Stop-phase session summaries, script-based hooks. Runtime controls viaECC_HOOK_PROFILE and ECC_DISABLED_HOOKS. New harness commands: /harness-audit, /loop-start, /loop-status, /quality-gate, /model-route. AgentShield integration for security scanning.
v1.6.0 — Codex CLI & AgentShield (Feb 2026)
Codex CLI support,/security-scan skill backed by AgentShield (1282 tests, 98% coverage, 102 static analysis rules), GitHub Marketplace launch, and 30+ community contributions.
Next Steps
Introduction
You are here — an overview of ECC, its primitives, supported harnesses, and version history.
Quickstart
Install ECC in five minutes, verify the setup, and run your first skill workflow.
Skills
Browse the skill library, understand the SKILL.md format, and invoke skills from your harness.
Hooks
Configure hook profiles, write custom hooks, and tune runtime enforcement behavior.