ECC: 68 Free Claude Code Agents in One Plugin (Install + Full List)

ECC is a plugin for Claude Code. One install gives Claude a crew of 68 specialist agents and 292 skills, plus hooks, rules and a security scanner. An agent is a helper with one job, like planning a feature, reviewing code or fixing a broken build. You don't have to pick them by hand. You ask in plain words and Claude hands the work to the right one.

It's free and open source under the MIT license. It works best with Claude Code, and the README also lists support for Codex and other coding tools.

Open Claude Code and paste these two lines, one at a time. They come straight from the ECC README:

Start a new Claude Code session so the plugin loads, then check it's on:

• Pick one install method. The README offers a guided setup and the two plugin commands above. Both install the same plugin. Don't stack them, or you get duplicate skills and hooks.
• It adds hooks. Hooks are small scripts that run on their own when Claude Code does things, like starting a session or editing a file. ECC's default hook profile is Standard. If you use the guided setup, choose Standard when it asks.
• It adds context at startup. ECC's session-start hook adds some notes to each new session (the README says up to 8,000 characters by default). That's part of what you pay for in plan usage.
• Uninstalling is easy. Run /plugin, open the Installed tab, pick ecc and choose Uninstall. Typing /plugin uninstall opens the same panel.
• Paste the repo URL into Claude first and ask what it will change on your computer. You get a plain-English summary before anything runs.

ECC is free, but running it is not free of cost. Claude Code's docs say each subagent runs in its own context window and sends its own requests, which count toward the same usage limits as your main chat. Anthropic's help center adds that Pro and Max limits are shared across Claude and Claude Code.

So every agent you call is extra Claude work on top of your own chat. On the Pro plan you will hit your limit sooner if you lean on agents a lot. If you run agents all day, a Max plan or pay-as-you-go API billing is the better fit. To see where your usage goes, type /usage in Claude Code. On Pro and Max it breaks usage down by skills, subagents and plugins.

You don't need to remember agent names. Say what you want and Claude picks the agent. Naming the agent works too.

• planner: say "plan this feature" and it maps out the steps before any code is written. You can also type /ecc:plan followed by the feature in quotes.
• code-reviewer: say "review it" after Claude writes code and it checks the changes for bugs, security and code quality. The shortcut is /ecc:code-review.
• build-error-resolver (bonus): say "fix this error" when your build or TypeScript check fails. It fixes the error with the smallest change it can, without redesigning your app. The shortcut is /ecc:build-fix.
• security-reviewer: say "use the security reviewer to check this project before I publish". It looks for hardcoded API keys, passwords and tokens, plus common holes like injection.

Straight from the agents folder in the repo. Short, plain versions of each agent's own description.

Planning and design (7)

• planner: plans complex features and refactors step by step.
• architect: system design, scaling and big technical decisions.
• code-architect: studies your codebase and writes a build blueprint with files, data flow and build order.
• code-explorer: traces how an existing feature works across your code before you change it.
• spec-miner: pulls the rules your existing code follows into written specs.
• type-design-analyzer: checks whether your types are well designed and hard to misuse.
• a11y-architect: accessibility (WCAG 2.2) for web and mobile UI.

Code review and cleanup (8)

• code-reviewer: reviews new code for quality, security and maintainability.
• code-simplifier: makes recent code clearer without changing what it does.
• comment-analyzer: checks code comments for accuracy and stale info.
• silent-failure-hunter: finds swallowed errors and bad fallbacks.
• pr-test-analyzer: checks whether a pull request's tests would really catch bugs.
• refactor-cleaner: finds and removes dead code and duplicates.
• performance-optimizer: finds slow code, memory leaks and bundle bloat.
• database-reviewer: PostgreSQL queries, schema design and migrations, with Supabase best practices.

Language and framework reviewers (17)

• typescript-reviewer: TypeScript and JavaScript code.
• react-reviewer: React components, hooks and rendering.
• vue-reviewer: Vue, Pinia, Vue Router and Nuxt code.
• python-reviewer: Python style, type hints, security and speed.
• django-reviewer: Django ORM, DRF and migration safety.
• fastapi-reviewer: FastAPI async code, Pydantic schemas and security.
• go-reviewer: idiomatic Go, concurrency and error handling.
• rust-reviewer: Rust ownership, lifetimes and unsafe code.
• java-reviewer: Java for Spring Boot and Quarkus projects.
• kotlin-reviewer: Kotlin, Android and Compose code.
• swift-reviewer: Swift, Swift Concurrency and memory management.
• flutter-reviewer: Flutter and Dart widgets and state management.
• cpp-reviewer: C++ memory safety and modern idioms.
• csharp-reviewer: C# and .NET conventions and async code.
• fsharp-reviewer: F# functional code and type safety.
• php-reviewer: PHP, PSR-12 and Eloquent patterns.
• harmonyos-app-resolver: HarmonyOS apps built with ArkTS and ArkUI.

Build fixers (11)

• build-error-resolver: fixes build and TypeScript errors with minimal changes.
• react-build-resolver: React build failures across Vite, webpack, Next.js and more.
• go-build-resolver: Go build and go vet errors.
• rust-build-resolver: cargo build and borrow checker errors.
• java-build-resolver: Java, Maven and Gradle build errors.
• kotlin-build-resolver: Kotlin and Gradle build errors.
• swift-build-resolver: Swift, Xcode and Swift Package Manager errors.
• cpp-build-resolver: C++, CMake and linker errors.
• dart-build-resolver: Dart and Flutter analysis, build and dependency errors.
• django-build-resolver: Django setup, migration and dependency errors.
• pytorch-build-resolver: PyTorch and CUDA training errors.

Security (1)

• security-reviewer: flags leaked secrets, injection, unsafe crypto and other OWASP Top 10 issues.

Testing (2)

• tdd-guide: test-driven development, tests first, aiming for 80%+ coverage.
• e2e-runner: writes and runs end-to-end tests of real user flows in a browser.

Docs (2)

• doc-updater: keeps READMEs, guides and code maps up to date.
• docs-lookup: fetches current library and API docs with examples.

AI, data and specialist reviewers (3)

• mle-reviewer: machine-learning pipelines, training, serving and monitoring.
• rag-pipeline-reviewer: RAG systems, chunking, embeddings and retrieval quality.
• healthcare-reviewer: healthcare code for clinical safety and patient-data compliance.

Agents that manage agents (7)

• agent-evaluator: scores an agent's output on accuracy, completeness, clarity and more.
• harness-optimizer: tunes your agent setup for reliability and cost using evals.
• loop-operator: runs long agent loops and steps in when they stall.
• conversation-analyzer: reads past chats to find habits worth blocking with hooks.
• gan-planner: turns a one-line idea into a full product spec.
• gan-generator: builds the spec and improves it from evaluator feedback.
• gan-evaluator: tests the running app in a browser and scores it.

Open-source release (3)

• opensource-forker: copies a project for open-sourcing and strips secrets and internal references.
• opensource-sanitizer: checks the copy for leaked secrets and personal data before release.
• opensource-packager: adds the README, license, setup script and contributing files.

Networks (4)

• network-architect: designs business or multi-site networks.
• network-config-reviewer: reviews router and switch configs for security and mistakes.
• network-troubleshooter: diagnoses connection, DNS and routing problems, read-only.
• homelab-architect: plans a home or small-lab network with safe, staged changes.

Business and growth (3)

• chief-of-staff: triages email, Slack and other messages and drafts replies.
• marketing-agent: campaign plans, landing pages, emails, ads and short-form scripts.
• seo-specialist: technical SEO audits, meta tags, schema and Core Web Vitals.

1. 1Paste the repo URL into Claude Code and ask what ECC will add to your setup.
2. 2Install it with the two official commands above. Nothing else.
3. 3Try the three reel agents on a small project: plan a feature, review it, run a security check. Then try the bonus one on your next error.
4. 4Check /usage after a day so you know what the agents cost you in plan usage.
5. 5Keep the agents you use. If it isn't for you, uninstall it from /plugin.