By 23blocks-OS
Run notification scripts on Claude Code stop and ping events, and execute context setup scripts at session start.
Authenticate AI agents with auth servers using the Agent Identity (AID) protocol. Supports Ed25519 identity documents, proof of possession, OAuth 2.0 token exchange, and scoped JWT tokens. Self-contained — works independently without other protocols.
Send and receive cryptographically signed messages between AI agents using the Agent Messaging Protocol (AMP). Supports local messaging, federation across providers, file attachments, and Ed25519 signatures. Works with any AI agent that can execute shell commands.
Creates, manages, and orchestrates AI agents using the AI Maestro CLI. Use when the user asks to "create agent", "list agents", "delete agent", "rename agent", "hibernate agent", "wake agent", "install plugin", "show agent", "export agent", "restart agent", "install marketplace", or any agent lifecycle management task.
Searches auto-generated codebase documentation for function signatures, API documentation, class definitions, and code comments. Use when the user asks to "search docs", "find function", "check API", or before implementing anything to understand existing patterns.
Queries the code graph database to understand component relationships, dependencies, and the impact of changes. Use when the user asks to "find callers", "check dependencies", "what uses this", or when exploring codebase structure to avoid breaking changes.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Build the exact set of skills your AI agent needs.
Remember when Neo sat in that chair and Tank, the operator, uploaded Kung Fu straight into his brain? Ten seconds later: "I know Kung Fu."
That's what this does for your Claude Code agents.
You pick the skills. You pick the scripts. You build a custom plugin. You install it. Your agent wakes up knowing everything you gave it.
A skill is a capability you load into your agent. Out of the box, AI Maestro comes with:
| Skill | What your agent can do | Source |
|---|---|---|
| Agent Messaging | Send and receive messages to other agents | agentmessaging/claude-plugin |
| Agent Identity | Passwordless OAuth 2.0 auth using Ed25519 keys | agentmessaging/agent-identity |
| Agent Management | Create, rename, hibernate, wake up other agents | Local (src/) |
| Memory Search | Search through past conversations and remember context | Local (src/) |
| Code Graph | Understand how your codebase is connected | Local (src/) |
| Docs Search | Search auto-generated documentation | Local (src/) |
| Planning | Break down complex tasks and track progress | Local (src/) |
The Source column tells you where each skill lives. Local means it's in this repo's src/ folder. Linked skills are pulled from external repos during build.
But here's the thing — you're not stuck with this list.
This repo is not a plugin you install as-is. It's a plugin builder.
Fork it. Edit one file. Pick the skills you want from anywhere — your own, your team's, some random genius on GitHub. The builder assembles them into one plugin, tailored to your agent.
Your agent. Your skills. Your call.
1. Fork this repo
2. Edit plugin.manifest.json ← pick your skills
3. Push (or run ./build-plugin.sh)
4. Install the plugin
5. Your agent knows Kung Fu
The manifest is your recipe. Want to add a code-review skill someone published on GitHub? Add three lines. Don't need memory search? Delete one line. Want to pull in your company's private deploy scripts? Add three lines.
CI builds automatically on push. You never run a build command if you don't want to.
# 1. Fork on GitHub, then:
git clone https://github.com/YOUR-USERNAME/ai-maestro-plugins.git
cd ai-maestro-plugins
# 2. (Optional) Edit plugin.manifest.json to customize
# 3. Build
./build-plugin.sh --clean
# 4. Install into Claude Code
claude plugin install ./plugins/ai-maestro
That's it. Your agent now has every skill in the manifest.
Say someone published an amazing code-review skill. Open plugin.manifest.json and add:
{
"name": "code-review",
"type": "git",
"repo": "https://github.com/alice/claude-skills.git",
"ref": "main",
"map": { "skills/code-review": "skills/code-review" }
}
Build. Install. Now your agent reviews code.
Delete the folder from src/skills/ and rebuild. Your plugin only includes what you keep.
Create a folder in src/skills/ with a SKILL.md file that tells Claude what to do. Push. CI builds. Done.
This is how every skill works — it's just a markdown file with instructions. No SDK. No framework. Just words that tell your agent what it's capable of.
Every AI agent is different because every project is different. A backend agent needs deploy scripts and database tools. A frontend agent needs design systems and accessibility checkers. A research agent needs web search and document analysis.
One-size-fits-all plugins don't work.
This builder lets you assemble exactly the right set of capabilities for each agent you run. Mix skills from open source repos, private team repos, and your own src/ folder — all into one plugin.
Same builder. Different agents. Different superpowers.
MIT
npx claudepluginhub 23blocks-os/ai-maestro-plugins --plugin ai-maestroUltra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Memory compression system for Claude Code - persist context across sessions
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills.