AGENTS-COLLECTION
Health Warn
- No license — Repository has no license file
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 50 GitHub stars
Code Fail
- fs.rmSync — Destructive file system operation in AGENTS/NANOCLAW/BASE/CONTAINER/AGENT-RUNNER/SRC/index.ts
- process.env — Environment variable access in AGENTS/NANOCLAW/BASE/CONTAINER/AGENT-RUNNER/SRC/index.ts
- process.env — Environment variable access in AGENTS/NANOCLAW/BASE/CONTAINER/AGENT-RUNNER/SRC/ipc-mcp-stdio.ts
Permissions Pass
- Permissions — No dangerous permissions requested
No AI report is available for this listing yet.
784 AGENT DEFINITIONS · 748 SKILLS · 141 PROMPTS — COLLECTED ACROSS 11 AI PLATFORMS: CLAUDE CODE, CURSOR, ANTIGRAVITY, OPENCODE, GITHUB COPILOT, NANOBOT, NANOCLAW, OPENCLAW, PICOCLAW, ZEROCLAW, AND MORE.
🚀 AGENTS-COLLECTION (Collective Production Edition)
💎 Overview
Fully production-grade implementation of AGENTS-COLLECTION, refactored by the 69-Agent Opencode Collective.
🛡️ Trust & Compliance
- CI/CD: Automated GitHub Actions with Gitleaks security scans.
- Security: Standardized SECURITY.md protocol.
- Design: Opencode Premium Design Tokens integrated.
🏁 48-Hour Roadmap
- Initialize infrastructure via
.github/workflows. - Set your secrets in GitHub Environment settings.
- Deploy to production via Vercel/Docker.
🤖 AGENTS-COLLECTION
The Most Comprehensive AI Agent Library on the Planet
A unified archive of AI agent definitions, skills, prompts, and configurations
collected, converted, and deployed across 11 AI coding platforms.
📖 TABLE OF CONTENTS
🧠 WHAT IS THIS?
This repository is a living snapshot of an AI-native development environment — every agent definition, skill file, prompt, workflow, MCP server config, hook, plugin, and platform config collected from a single power-user's machine across 11 AI coding assistants.
Core value propositions:
- 🔍 Discover — Browse 700+ unique agent definitions in one place
- 🔄 Cross-platform — See how the same agent is expressed in 6+ different formats
- 🚀 Deploy — Copy any folder directly into your own tool's config directory
- 📚 Learn — Study agent design patterns, skill structures, and prompt engineering at scale
🖥️ PLATFORM COVERAGE
| Platform | Type | Agents/Skills | Source Folder | Global Config Path |
|---|---|---|---|---|
| Claude Code | AI CLI | 129 agents | AGENTS/CLAUDE-CODE/ |
~/.claude/agents/ |
| AntiGravity (Gemini) | AI CLI | 82 skills | AGENTS/ANTIGRAVITY/SKILLS/ |
~/.gemini/antigravity/skills/ |
| Cursor | AI IDE | 68 rules | AGENTS/CURSOR/ |
~/.cursor/rules/ |
| OpenCode | AI Editor | 68 agents | AGENTS/OPENCODE/ |
~/.config/opencode/agent/ |
| GitHub Copilot | AI Pair | 68 agents | AGENTS/PICOCLAW/ |
~/.github/copilot-instructions.md |
| NanoBot | Agent Runner | 68 skills | SKILLS/NANOBOT/ |
~/.nanobot/skills/ |
| NanoClaw (Claude) | Agent Platform | 89 skills | SKILLS/NANOCLAW/ |
.claude/skills/ |
| OpenClaw | Multi-Agent | 22 agents | AGENTS/OPENCLAW/ |
.openclaw/agents/ |
| PicoClaw (Copilot) | Code Agent | 2 configs | AGENTS/PICOCLAW/ |
.github/copilot-instructions.md |
| ZeroClaw (Gemini) | Code Agent | 69 prompts | PROMPTS/ZEROCLAW-GEMINI/ |
.gemini/agents/ |
| Everything-CC | Plugin Pack | 13 agents | AGENTS/EVERYTHING-CC/ |
~/.claude/agents/ |
📁 REPOSITORY STRUCTURE
AGENTS-COLLECTION/
│
├── AGENTS/ ← Agent definitions by platform
│ ├── CLAUDE-CODE/ │ 129 .md files — global Claude Code agents
│ ├── ANTIGRAVITY/ │ AntiGravity (Gemini CLI) data
│ │ ├── SKILLS/ │ 82 skill folders — each with SKILL.md
│ │ ├── ANNOTATIONS/ │ Code annotation data
│ │ ├── RULES/ │ AntiGravity rule files (.pbtxt)
│ │ └── WORKFLOWS/ │ Workflow definitions
│ ├── CURSOR/ │ 68 .mdc cursor rule files
│ ├── OPENCODE/ │ 68 .md OpenCode agent definitions
│ ├── PICOCLAW/ │ GitHub Copilot instruction files
│ ├── NANOCLAW/ │ 41 NanoClaw agent files
│ ├── OPENCLAW/ │ 22 multi-agent team configs
│ │ ├── FEATURE-DEV_*/ │ 6-agent feature dev team
│ │ ├── BUG-FIX_*/ │ 6-agent bug fix team
│ │ └── SECURITY-AUDIT_*/ │ 7-agent security team
│ ├── EVERYTHING-CC/ │ 13 Everything-Claude-Code agents
│ └── AGENCY-SOURCE/ │ 85 source agents (canonical MD)
│ ├── DESIGN/ │ 7 design agents
│ ├── ENGINEERING/ │ 11 engineering agents
│ ├── MARKETING/ │ 8 marketing agents
│ ├── PRODUCT/ │ 6 product agents
│ ├── PROJECT-MANAGEMENT/ │ 5 PM agents
│ ├── SPECIALIZED/ │ 14 specialized agents
│ ├── SPATIAL-COMPUTING/ │ 5 XR/spatial agents
│ ├── SUPPORT/ │ 5 support agents
│ ├── TESTING/ │ 6 testing agents
│ └── STRATEGY/ │ extra strategy agents
│
├── SKILLS/ ← Installable skills & capabilities
│ ├── CLAUDE-CODE/ │ 499 files — SKILL.md + manifests
│ ├── NANOCLAW/ │ 262 files — 89 containerized skills
│ ├── OPENCLAW/ │
│ │ ├── SRC/ │ 27 files — global OpenClaw skills
│ │ └── PROJECT/ │ 74 files — project-level skills
│ ├── NANOBOT/ │ 69 folders — skill.json + SKILL.md
│ └── EVERYTHING-CC/ │ 48 files + 51 OpenCode skill configs
│ └── OPENCODE/ │
│
├── PROMPTS/ ← Prompt collections
│ ├── OPENCLAW-PI/ │ 72 Pi prompts (frontmatter .md)
│ └── ZEROCLAW-GEMINI/ │ 69 Gemini Code Assist agent files
│
├── PLUGINS/ ← Plugin manifests & extensions
│ ├── CLAUDE-CODE/ │ 500 files — manifests + docs
│ └── OPENCLAW/ │ 9 OpenClaw extension files
│
├── WORKFLOWS/ ← Workflow definitions & plans
│ ├── CLAUDE-PLANS/ │ 8 Claude Code implementation plans
│ └── OPENCLAW/WORKFLOWS/ │ 1 OpenClaw workflow definition
│
├── MCP/ ← MCP server configs (secrets redacted)
│ ├── CLAUDE-CODE/ │ mcp.json + .mcp.json
│ ├── NANOCLAW/ │ .mcp.json
│ ├── CURSOR/ │ mcp.json
│ └── OPENCLAW/ │ MCP configs
│
├── HOOKS/ ← Pre/post execution hooks
│ ├── CLAUDE-CODE/ │ Claude Code hooks
│ └── OPENCLAW/ │ OpenClaw hooks
│
├── RULES/ ← Rule files
│ └── CLAUDE-CODE/ │ .md rule files (coding style, git, etc.)
│
├── DOCS/ ← Documentation
│ ├── NANOCLAW/ │ Architecture docs
│ ├── OPENCLAW/ │ AGENTS.md, CLAUDE.md
│ ├── ZEROCLAW/ │ AGENTS.md, CLAUDE.md
│ └── PICOCLAW/ │ 19 doc files
│
└── CONFIGS/ ← Platform configs (secrets redacted)
├── CLAUDE-CODE/ │ GEMINI.md
├── OPENCLAW/ │ openclaw.json
├── NANOBOT/ │ config.json
├── CURSOR/ │ mcp.json, cli-config.json
├── PICOCLAW/ │ config.json
└── ZEROCLAW/ │ config.toml, zeroclaw.json
📊 FILE TYPE BREAKDOWN
| Extension | Count | Approx Size | Category | Used By |
|---|---|---|---|---|
.md |
2,186 | ~8 MB | Agent defs, skills, docs, prompts | All platforms |
.json |
168 | ~3 MB | Configs, manifests, skill.json | NanoBot, NanoClaw, Plugins |
.ts |
122 | ~1.5 MB | TypeScript skill implementations | NanoClaw, Everything-CC |
.yaml |
80 | ~400 KB | Manifests, NanoClaw configs | NanoClaw, OpenClaw |
.mdc |
68 | ~1.5 MB | Cursor rule files | Cursor |
.pbtxt |
19 | ~200 KB | AntiGravity rule/workflow data | AntiGravity |
.py |
16 | ~200 KB | Python skill scripts | NanoClaw |
.txt |
14 | ~100 KB | Text prompts, notes | Various |
.mjs |
13 | ~150 KB | JavaScript modules | Everything-CC |
.sh |
11 | ~50 KB | Shell scripts, installers | Various |
.toml |
1 | ~2 KB | ZeroClaw config | ZeroClaw |
Dockerfile |
2 | ~3 KB | Container skill definitions | NanoClaw |
Markdown dominates (80%) — all platforms converge on
.mdas the universal agent definition format.
Excluded from this repo: 600+ MB of runtime data (.pb conversation files, sessions.json, brain state, code tracking) — only agent definitions are committed.
🤖 AGENTS
Claude Code — 129 Agents
Global agents at ~/.claude/agents/ covering the full development lifecycle:
| Category | Agents |
|---|---|
| Engineering | backend-architect, frontend-developer, mobile-app-builder, devops-automator, rapid-prototyper, ai-engineer, security-engineer, data-engineer, senior-developer, technical-writer, autonomous-optimization-architect |
| Design | ui-designer, ux-architect, brand-guardian, visual-storyteller, image-prompt-engineer, inclusive-visuals-specialist, whimsy-injector |
| Marketing | content-creator, tiktok-strategist, trend-researcher, app-store-optimizer, community-manager, analytics-reporter, campaign-manager, growth-hacker |
| Product | product-writer, sprint-prioritizer, feedback-synthesizer, studio-producer, experiment-tracker, project-shipper |
| Testing | test-writer-fixer, api-tester, qa-specialist, test-results-analyzer |
| Support | customer-support, onboarding-guide, documentation-specialist, knowledge-base-curator, feedback-analyst |
| Specialized | joker, nexus-strategy, tool-evaluator, zhihu-strategist, and 40+ more |
| System (Agency) | planner, architect, tdd-guide, code-reviewer, security-reviewer, build-error-resolver, e2e-runner, refactor-cleaner, doc-updater |
AntiGravity (Gemini CLI) — 82 Skills
Installed at ~/.gemini/antigravity/skills/. Each skill is a named folder with SKILL.md:
AGENCY-BACKEND-ARCHITECT/
└── SKILL.md # name, description, risk_level, source frontmatter
Includes all 68 Agency agents + 14 additional GWS recipe skills.
Agency Source — 68 Canonical Agents
Original msitarzewski/agency-agents definitions in AGENTS/AGENCY-SOURCE/:
| Division | Count | Sample Agents |
|---|---|---|
| Engineering | 11 | backend-architect, frontend-developer, devops-automator, ai-engineer |
| Design | 7 | ui-designer, ux-architect, brand-guardian, whimsy-injector |
| Marketing | 8 | content-creator, tiktok-strategist, trend-researcher, app-store-optimizer |
| Product | 6 | product-writer, sprint-prioritizer, feedback-synthesizer, studio-producer |
| Project Mgmt | 5 | studio-coach, experiment-tracker, project-shipper |
| Testing | 6 | test-writer-fixer, api-tester, qa-specialist, e2e-runner |
| Support | 5 | customer-support, documentation-specialist, knowledge-base-curator |
| Spatial Computing | 5 | xr-developer, ar-experience-designer, vr-environment-builder |
| Specialized | 14 | joker, nexus-strategy, tool-evaluator, zhihu-strategist + more |
OpenClaw Multi-Agent Teams — 22 Configs
AGENTS/OPENCLAW/ — Production multi-agent team configs:
| Team | Agents | Purpose |
|---|---|---|
| feature-dev | 6 | planner → developer → reviewer → tester → verifier → setup |
| bug-fix | 6 | triager → investigator → fixer → pr → setup → verifier |
| security-audit | 7 | scanner → prioritizer → fixer → tester → verifier → pr → setup |
| acp / codex / main | 3 | Orchestration & coordination |
⚡ SKILLS
Claude Code Skills — 499 Files
SKILLS/CLAUDE-CODE/ — The most comprehensive Claude Code skill library:
| Category | Count | Description |
|---|---|---|
| GWS Skills | 80+ | Gmail, Drive, Calendar, Sheets, Docs, Slides, Meet, Chat, Classroom, Forms |
| Persona Skills | 10 | exec-assistant, researcher, project-manager, customer-support, sales-ops |
| Recipe Skills | 50+ | batch-invite, backup-sheet, compare-tabs, draft-from-doc, find-free-time |
| Agency Skills | 68 | All 68 Agency agents as installable skills |
| Code Skills | 20+ | TDD, review, security, refactor, build-error-resolver, e2e |
| Everything-CC | 48 | Battle-tested skills from Everything-Claude-Code |
NanoClaw Skills — 262 Files (89 Skill Folders)
SKILLS/NANOCLAW/ — Docker-based containerized skills:
SKILL-NAME/
├── SKILL.md # Description + manifest
├── manifest.yaml # NanoClaw manifest
└── MODIFY/
├── Dockerfile # Container definition
└── SRC/ # Implementation
Notable skills: ADD-PDF-READER, USE-LOCAL-WHISPER, CONVERT-TO-APPLE-CONTAINER, SHERPA-ONNX-TTS, ADK-TOOL-SCAFFOLD + 84 more.
NanoBot Skills — 69 Folders
SKILLS/NANOBOT/ — Each folder: skill.json + SKILL.md:
{
"name": "agency-backend-architect",
"description": "Expert backend systems designer",
"trigger": "design API|architect backend|scale service",
"model": "claude-sonnet-4-5"
}
Everything-CC Skills — 99 Files
SKILLS/EVERYTHING-CC/ — From disler/everything-claude-code:
- TDD workflow, code review, security scan, python/go/java review
- Django, SpringBoot, FastAPI, PostgreSQL patterns
- Deployment, Docker, E2E testing, ClickHouse patterns
- Content-hash-cache, eval-harness, iterative-retrieval
💬 PROMPTS
OpenClaw Pi Prompts — 72 Files
PROMPTS/OPENCLAW-PI/ — Frontmatter-based prompts for the Pi intelligence system:
---
name: backend-architect
description: Design scalable backend systems
category: engineering
---
[prompt body]
ZeroClaw Gemini Prompts — 69 Files
PROMPTS/ZEROCLAW-GEMINI/ — Gemini Code Assist compatible agent files:
# Backend Architect
> Design scalable APIs and distributed systems
[full agent prompt]
🔌 PLUGINS · MCP · HOOKS · CONFIGS
Claude Code Plugins — 500 Files
PLUGINS/CLAUDE-CODE/ — Full plugin manifests + multilingual documentation.
Includes EVERYTHING-CLAUDE-CODE plugin cache with 30+ skills in English, Chinese (zh-cn), and Japanese (ja-jp).
MCP Server Configs (secrets redacted)
MCP/ — Model Context Protocol server configurations:
| File | MCP Servers Configured |
|---|---|
CLAUDE-CODE/mcp.json |
filesystem, github, figma, supabase, vercel, railway, cloudflare, firecrawl, memory, sequential-thinking, playwright, web-search, linear, gmail, canva, gamma, indeed, zread, context7 |
CLAUDE-CODE/.mcp.json |
Project-level overrides |
CURSOR/mcp.json |
Cursor MCP (same 20+ servers) |
NANOCLAW/.mcp.json |
NanoClaw project MCP |
⚠️ All API tokens, PATs, and secrets replaced with
***REDACTED***before commit.
Hooks — 8 Files
HOOKS/ — Pre/post execution lifecycle hooks:
- Session-start hooks (context loading, memory init)
- Tool-use hooks (validation, auto-format)
- Stop hooks (final verification, cleanup)
Configs (secrets redacted)
CONFIGS/ — Platform configuration files:
| File | Platform | Notes |
|---|---|---|
CLAUDE-CODE/GEMINI.md |
Gemini integration | Gemini model config |
OPENCLAW/openclaw.json |
OpenClaw platform | Agent team definitions |
NANOBOT/config.json |
NanoBot | 20+ LLM provider configs |
CURSOR/mcp.json |
Cursor | MCP + tool config |
ZEROCLAW/config.toml |
ZeroClaw | Gemini CLI config |
PICOCLAW/config.json |
PicoClaw | Copilot config |
🌟 THE 68 CORE AGENCY AGENTS
All 68 agents from msitarzewski/agency-agents installed across 5 tools simultaneously in their native formats:
| Agent | Claude Code | Cursor | OpenCode | AntiGravity | Copilot |
|---|---|---|---|---|---|
| backend-architect | ✅ .md |
✅ .mdc |
✅ .md |
✅ SKILL.md |
✅ concat |
| frontend-developer | ✅ | ✅ | ✅ | ✅ | ✅ |
| mobile-app-builder | ✅ | ✅ | ✅ | ✅ | ✅ |
| devops-automator | ✅ | ✅ | ✅ | ✅ | ✅ |
| rapid-prototyper | ✅ | ✅ | ✅ | ✅ | ✅ |
| ai-engineer | ✅ | ✅ | ✅ | ✅ | ✅ |
| security-engineer | ✅ | ✅ | ✅ | ✅ | ✅ |
| ui-designer | ✅ | ✅ | ✅ | ✅ | ✅ |
| ux-architect | ✅ | ✅ | ✅ | ✅ | ✅ |
| content-creator | ✅ | ✅ | ✅ | ✅ | ✅ |
| tiktok-strategist | ✅ | ✅ | ✅ | ✅ | ✅ |
| product-writer | ✅ | ✅ | ✅ | ✅ | ✅ |
| ... (56 more) | ✅ | ✅ | ✅ | ✅ | ✅ |
Also installed in NanoBot, NanoClaw, OpenClaw, PicoClaw, ZeroClaw — 10 platforms total.
Each format differs:
| Platform | File Format | Key Fields |
|---|---|---|
| Claude Code | .md + YAML frontmatter |
name, description, model, tools |
| Cursor | .mdc |
description, globs, alwaysApply |
| OpenCode | .md + frontmatter |
name, description, color |
| AntiGravity | folder/SKILL.md |
name, description, risk_level, source |
| GitHub Copilot | concat in single .md |
plain markdown sections |
| NanoBot | skill.json + SKILL.md |
name, description, trigger, model |
| NanoClaw | manifest.yaml + SKILL.md |
name, version, description |
| OpenClaw Pi | frontmatter .md |
name, description, category |
| ZeroClaw | plain .md |
# Title + > description |
🚀 HOW TO USE
Install — Claude Code Agents
cp AGENTS/CLAUDE-CODE/*.md ~/.claude/agents/
ls ~/.claude/agents/ | wc -l # verify
Install — AntiGravity Skills
cp -r AGENTS/ANTIGRAVITY/SKILLS/* ~/.gemini/antigravity/skills/
Install — Cursor Rules
cp AGENTS/CURSOR/*.mdc ~/.cursor/rules/
Install — OpenCode Agents
cp AGENTS/OPENCODE/*.md ~/.config/opencode/agent/
Install — NanoBot Skills
cp -r SKILLS/NANOBOT/* ~/.nanobot/skills/
Install — NanoClaw Skills
# Per-project install
cp -r SKILLS/NANOCLAW/ADD-PDF-READER /path/to/project/.claude/skills/
Browse & Search
# Find agents by topic
grep -rl "security" AGENTS/ --include="*.md" | head -20
# Find skills with specific capability
grep -rl "TypeScript" SKILLS/ --include="*.md" | head -10
# List all agents by platform
ls AGENTS/CLAUDE-CODE/
ls AGENTS/OPENCODE/
ls AGENTS/CURSOR/
# Find duplicate agent definitions across platforms
grep -rh "^name:" AGENTS/ --include="*.md" | sort | uniq -d
🔗 REFERENCES & SOURCE REPOS
Source Projects
| Project | Author | Link | Contribution |
|---|---|---|---|
| agency-agents | msitarzewski | github.com/msitarzewski/agency-agents | 68 canonical agent definitions — the backbone |
| everything-claude-code | disler | github.com/disler/everything-claude-code | 48 skills, 13 agents, OpenCode configs |
| gemini-cli | github.com/google-gemini/gemini-cli | AntiGravity skill platform | |
| NanoClaw | mk-knight23 | github.com/mk-knight23/NanoClaw | Containerized Claude skill platform |
| openclaw | mk-knight23 | github.com/mk-knight23/openclaw | Multi-agent team platform |
| zeroclaw | mk-knight23 | github.com/mk-knight23/zeroclaw | Gemini-based code agent |
| picoclaw | mk-knight23 | github.com/mk-knight23/picoclaw | Copilot-based Go code agent |
Platform Docs
| Platform | Documentation |
|---|---|
| Claude Code agents | docs.anthropic.com/claude-code |
| Gemini AntiGravity | github.com/google-gemini/gemini-cli |
| Cursor rules | docs.cursor.com/context/rules |
| OpenCode agents | opencode.ai |
| GitHub Copilot instructions | docs.github.com/copilot/customizing-copilot |
| MCP (Model Context Protocol) | modelcontextprotocol.io |
🙏 CREDITS & CONTRIBUTIONS
Primary Creator
Musharraf Kazi — @mk-knight23
AI Organization Architect · 6+ years AI engineering · Building self-evolving digital organizations
Current projects: Open-Universe (60 apps), AI-VIBE Ecosystem, AI-SDK-Ecosystem, RALPH (134+ agents, 21 teams)
Agency Agents — Original Author
Mike Sitarzewski — @msitarzewski
Creator of agency-agents — the 68-agent personality system and the
convert.shscript that generates 6 platform formats from a single source. All 68 core agents originate from his work. ⭐ Star his repo.
Everything Claude Code — Original Author
Daniel Isler — @disler
Creator of everything-claude-code — 48+ production-grade skills, coding standards, and agent configurations. ⭐ Star his repo.
Platform Teams
| Team | Contribution |
|---|---|
| Anthropic | Claude Code agent system, MCP protocol specification |
| Google DeepMind | Gemini CLI + AntiGravity skill platform |
| Cursor | .mdc rule system for AI-native IDE |
| OpenCode Labs | OpenCode agent format and global agent directory |
| GitHub | Copilot custom instructions system |
📋 COLLECTION STATS
| Metric | Value |
|---|---|
| Total Size (definitions) | ~30 MB |
| Total Size (with runtime, excluded) | ~600 MB |
| Total Files (committed) | 2,700+ |
| Total Directories | 1,300+ |
| Platforms Covered | 11 |
| Source Locations Scanned | 15 |
| Markdown Files | 2,186 |
| JSON Configs | 168 |
| TypeScript Files | 122 |
| YAML Manifests | 80 |
| Cursor Rules (.mdc) | 68 |
| Shell Scripts | 11 |
| Removed (runtime) | ~499 files — conversation/session history |
| Removed (duplicates) | ~340 files — old lowercase paths + integration copies |
| Secrets Redacted | 6 files (MCP configs, platform configs) |
📄 LICENSE
MIT — All original work by @mk-knight23
Individual agent definitions retain their original licenses:
Built with obsession. Deployed across 11 platforms. Runtime excluded. Definitions preserved.
⭐ Star this repo · 🍴 Fork it · 📦 Use the agents
Security
This project follows security best practices:
- No hardcoded credentials
- Dependency scanning enabled
- Security headers configured
- Regular security audits performed
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found