Everything Claude Code (ECC)

Reviewed 2026-10-15agentsPrompt / Rule# agents # claude-code # rules # hooks # skills

A large open-source toolkit of agent definitions, slash commands, rules, hooks and skills for coding agents. Built for Claude Code; it also has a native Antigravity installer that copies rules, workflows, skills and agents into .agents/.

Security Reviewed

Mostly Markdown; Antigravity install adds no hooks. Claude Code hooks run Node scripts; some skills include scripts.

Install Everything Claude Code (ECC) in Google Antigravity

Antigravity loads skills from .agents/skills/ in your project or ~/.gemini/config/skills/ for all projects (official docs). Full guide

EasiestOption 1: paste this to your Antigravity agent

Install the "Everything Claude Code (ECC)" skill for Google Antigravity.
Follow the steps at https://www.antigravityskills.org/install/everything-claude-code.md (source: https://github.com/affaan-m/everything-claude-code).
Before running anything, show me what you will download and whether it includes scripts, and wait for my OK.
Install it into this project (.agents/skills/) unless I say global. When done, check that Antigravity lists it as an installed skill.

The agent reads the install file, shows you what it will download and waits for your OK.

Option 2: install it yourself

1. Check Node.js (18 or newer)
```
node --version
```
The README requires Node.js 18 or newer for the universal package.
2. See what would be installed first
```
npx ecc-universal@2.2.3 install --profile minimal --target antigravity --dry-run
```
Run from the project you want to configure. docs/ANTIGRAVITY-GUIDE.md uses ecc-universal@2.2.0 or newer for the native layout; we reviewed repository version 2.2.3 (VERSION file). scripts/install-apply.js documents a --dry-run flag in its usage text; use it to see the plan before anything is written.
3. Install the minimal profile for Antigravity
```
npx ecc-universal@2.2.3 install --profile minimal --target antigravity
```
Writes into .agents/ in the current project: rules/, workflows/ (from commands), skills/, agents/ and ecc-install-state.json. This is the project-local layout; the guide also has a source-checkout alternative with ./install.sh --profile minimal --target antigravity.
4. Start a new conversation and verify
```
rg --files .agents/skills -g 'SKILL.md'
```
In Antigravity open Settings > Customizations, confirm the workspace skills appear, start a new conversation and ask for one by exact name (for example tdd-workflow). The guide also lists node scripts/doctor.js --target antigravity from a source checkout.

Steps follow docs/ANTIGRAVITY-GUIDE.md and the README at b92f181; we have not run the installer or tested it in Antigravity. To remove it later use the uninstall script described in the README's Reset / Uninstall section.

Using Claude Code, Cursor or Codex? See other assistants below.

What is Everything Claude Code (ECC)?

Everything Claude Code, now named ECC and published on npm as ecc-universal, is a big collection of files for coding agents rather than a single skill. At the reviewed commit the repository holds 71 agent definitions, 97 slash commands, 311 skill folders and 122 rule files, plus a Node.js installer with 'profiles' that choose which parts to copy. Claude Code is its primary target. For Antigravity 2.0 it has a native target that writes into your project's .agents/ folder. The pieces cover test-driven development, code review, security checks, planning and session memory.

How to use Everything Claude Code (ECC)?

Pick a profile and a target, run the installer from the project you want to configure, then start a new conversation so the agent sees the new skills. The README's low-context path is the 'minimal' profile, which excludes the hook runtime. Ask for a workflow by name, for example the tdd-workflow skill, or run one of the installed commands as a workflow.

Key features of Everything Claude Code (ECC)

✓We counted at b92f181: 71 files in agents/, 97 in commands/, 311 folders in skills/ and 122 files under rules/
✓We counted: manifests/install-modules.json defines 46 install modules and manifests/install-profiles.json 7 profiles (minimal, opencode, core, developer, security, research, full)
✓We counted: the minimal profile selects 52 skill folders for the Antigravity target, plus the rules, agents and commands (Antigravity workflows) - not all 311 skills
✓The Antigravity target (scripts/lib/install-targets/antigravity-project.js) copies only rules, commands, agents and skills into .agents/ and writes .agents/ecc-install-state.json; the hooks-runtime module does not list antigravity as a target
✓The Claude Code hook set in hooks/hooks.json has 27 entries (10 PreToolUse, 8 Stop, 2 SessionStart, 2 PostToolUse, 2 PostToolUseFailure, 1 each UserPromptSubmit, PreCompact, SessionEnd) that run Node scripts when you use Claude Code with the hook runtime installed
✓Installers refuse to write hooks until you pass --enable-hooks or --no-hooks (README, 'Low-context / no-hooks path')

When to use Everything Claude Code (ECC)

1Adding a test-first workflow (tdd-workflow, verification-loop) to an Antigravity or Claude Code project
2Giving the agent project rules for a language (rules/ has common and per-language files)
3Trying specialised review agents such as code-reviewer on a codebase
4Teams that want the same rules and agents across several coding harnesses

Installation & Setup Guide

Agent Rules / Prompt Setup

Follow these steps (taken from the project's own docs) to configure Everything Claude Code (ECC) in your development environment.

Step 1: Claude Code (guided setup)claude

npx ecc-universal@2.2.3 setup

The README's recommended path for Claude Code plugin setup; it needs Git and Claude Code 2.1 or newer on PATH.

Step 2: Claude Code without hooksclaude

npx ecc-universal@2.2.3 install --profile minimal --target claude

The README's low-context path; this profile excludes the hook runtime.

Step 3: Cursorcursor

./install.sh --profile minimal --target cursor

From a clone of the repository, per the README table (project-local .cursor/ adapter, listed as a beta adapter).

Step 4: Gemini CLIgemini

./install.sh --profile minimal --target gemini

From a clone of the repository, per the README table (project-local .gemini/ config, experimental adapter).

Usage Example & Verification

Prompt / Request Example

1. (our example)
Use the tdd-workflow skill: add a function that parses ISO dates in utils/date.ts. Write the failing test first, then the code, then show me the test results.
-> The agent should follow the skill's red-green-refactor steps rather than writing code first.

2. (our example)
Use the verification-loop skill on my last change and list what it checked.
-> A 135-line SKILL.md at the reviewed commit; expect build, lint and test checks if your project has them.

3. (README style)
Ask for the code-reviewer agent on a diff. Its definition declares tools Read, Grep, Glob and Bash, so it can run shell commands during review.

Security Review

Static review

Reviewed Date

2026-10-15

Audited Ref / Commit

b92f181b3b748d1abbc2afd94af6aa7d8c8fca29

Reviewer

antigravityskills.org (static review)

Mostly Markdown; Antigravity install adds no hooks. Claude Code hooks run Node scripts; some skills include scripts.

Full review: scope, findings, risks

Audit Scope:README.md (install, profiles, platform support, hook sections), docs/ANTIGRAVITY-GUIDE.md, package.json, VERSION, SECURITY.md, LICENSE, hooks/hooks.json (all entries listed), manifests/install-modules.json and install-profiles.json, scripts/lib/install-targets/antigravity-project.js, the network-capable hook scripts mcp-health-check.js and plan-canvas-pending.js (matched by grep for http/fetch, headers read), skills/continuous-learning-v2/SKILL.md, agents/code-reviewer.md, plus directory counts, all at commit b92f181 (2026-10-11). Static review only: the installer was not run, and the 4,615 files were not read line by line.

Key Findings:Large instruction collection with executable parts. For the Antigravity target the planner copies only rules, commands, agents and skills and never the hooks runtime. For Claude Code, hooks/hooks.json registers 27 hook entries that run Node. Two hook scripts import an http client (mcp-health-check.js with https, plan-canvas-pending.js with http); we did not trace what they contact. continuous-learning-v2 stores observer data under ~/.local/share/ecc-homunculus (or XDG data home) and its background observer is off by default (observer.enabled false in its table). The README includes affiliate links (aff=ecc) for a model provider and a sponsor section.

Identified Operational Risks & Permissions:Medium for the full Claude Code hook setup, low for the minimal Antigravity install of text files. The size of the catalog is the main practical risk: a bloated context and agents following many instructions. Do not enable hooks or run optional Python/shell skills without reading them first.

We read the files listed under review scope. The Antigravity adapter source filters module paths to rules, commands, agents and skills. hooks-runtime lists claude, claude-project, cursor, opencode and codebuddy as targets, not antigravity. hooks.json commands are node -e launchers that resolve the plugin root and run scripts from the package. Network-capable hooks and scripts exist (two hook scripts import http/https) and we did not audit their destinations, so we do not claim the hook runtime is offline. agents/code-reviewer.md declares Bash among its tools. SECURITY.md points to GitHub private vulnerability reporting. This is a partial static read, not a full audit of 4,615 files.

Operational Notes & Known Limitations

Known Limitations

•It is a very large toolkit (4,615 files at the reviewed commit); the README itself warns that the Claude plugin advertises the whole catalog to the model, so choose a selective profile when context size matters
•The README rates Antigravity as an 'experimental/minimal adapter': file placement and instruction portability are tested, full Claude Code feature parity is not claimed
•Hooks, the observer daemon and several shell-backed skills are for Claude Code and need Node, Bash or Python; none of them is installed by the Antigravity target
•The npm/installer path runs code from the package on your machine; read the profile and dry-run output before applying it
•The repository was renamed (package.json points to affaan-m/ECC); the old everything-claude-code URL redirects there

Frequently Asked Questions

Is Everything Claude Code safe to install?

The repository is MIT licensed and its README documents what each profile installs. The Antigravity target only copies Markdown rule, command, agent and skill files into .agents/. The Claude Code hook runtime runs Node scripts on your tool calls and some skills include Python or shell scripts, so use the minimal profile first and read what you install.

Does it work in Antigravity?

Yes, through a native target for Antigravity 2.0 (ECC 2.2.0 or newer), but the README labels it experimental: files are placed and the instructions are portable, while Claude Code hooks and full feature parity are not part of it.

How many skills does the Antigravity install add?

We counted 52 skill folders in the minimal profile's modules for Antigravity, out of 311 skill folders in the repository. The same install also brings the rules (122 files in the repository), the agent definitions (71) and the commands, which Antigravity receives as workflows (97).

Will it slow down or crowd my agent?

It can. The README warns that the plugin lists the whole catalog to the model and suggests selective profiles; install only what you use.

Is it the same as the ECC or ecc-universal package?

Yes. The project is now named ECC (npm: ecc-universal) and the repository URL redirects from everything-claude-code.

Supported Platforms

claudeantigravitycursorgemini

View Source on GitHub

Keep exploring this topic

Browse the full Antigravity skills directory for more source-linked skills with install steps and static safety reviews.

More agents skills #agents #claude-code #rules #hooks Use cases

Related Skills

💻

TDD Workflow

@sickn33

Community skill that makes an agent work test-first: one failing behavioural test, minimal code to pass, then refactor, with a separate guide on mock anti-patterns.

Static reviewinstructions onlyoffline

tddtesting

🛡️

Secrets Management

@sickn33

Community skill with a short procedure for handling CI/CD secrets safely: Vault, cloud secret managers, GitHub Actions, workload identity and rotation.

Static review

secretsci-cd

💻

Software Architecture

@sickn33

Community skill that gives an agent a simplicity-first method for architecture decisions: ask context questions, pick patterns with decision trees, and record the choice as an ADR.

Static reviewinstructions onlyoffline

architectureadr