{"id":"d35f714d-3828-40e3-8912-af4c92a8288b","slug":"hanzoai-mcp","kind":"mcp","name":"mcp","summary":"Model Context Protocol server with 260+ tools for AI agents","owner":"hanzoai","repoUrl":"https://github.com/hanzoai/mcp","license":"MIT","licenseTier":"permissive","trust":"community","stars":9,"match":null,"fewStars":false,"format":"mcp","createdAt":"2025-03-24T03:19:32.000Z","addedAt":"2026-10-14T05:27:51.027Z","logoUrl":"https://avatars.githubusercontent.com/u/16248274?v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["memory-knowledge","cloud-infra"],"tags":["ai-agents","ai-infrastructure","llm-tools"],"runtime":"node","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-26T14:09:04.466Z","sample":false,"url":"https://conduits.dev/tools/hanzoai-mcp","language":"Rust","transports":["stdio"],"install":{"npm":"@hanzo/mcp"},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"<p align=\"center\"><img src=\".github/hero.svg\" alt=\"Hanzo MCP\" width=\"880\"></p>\n\n# Hanzo MCP\n\n[![npm](https://img.shields.io/npm/v/hanzo?color=blue&label=hanzo)](https://www.npmjs.com/package/hanzo)\n[![Protocol](https://img.shields.io/badge/MCP%20Protocol-2025--06--18-purple.svg)](https://modelcontextprotocol.io)\n[![License](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)\n[![Gateway](https://img.shields.io/badge/API-api.hanzo.ai-orange.svg)](https://api.hanzo.ai)\n\nThe official **Model Context Protocol (MCP)** server for coding agents, autonomous workflows, and tool orchestration. \n\nIt exposes 260+ capabilities collapsed into high-efficiency action-routed tools over **stdio** or **HTTP/SSE** (`hanzo mcp -t sse`), plus native integration with the **Kai Agent Governance & Decision Engine**.\n\n---\n\n## Install\n\nOne install, any OS (Windows included):\n\n```bash\nnpm i -g hanzo                      # or, on macOS/Linux: curl -fsSL https://hanzo.sh | sh\nhanzo auth login                    # sign in\n```\n\n`npm i -g hanzo` installs a small launcher. Each native Rust binary is downloaded on\nfirst use and its SHA-256 checked:\n\n| Command | What it runs |\n| :--- | :--- |\n| `hanzo mcp` | this MCP server (native Rust `hanzo-mcp`) over stdio; `hanzo mcp -t sse` serves HTTP/SSE |\n| `hanzo dev` | the [Hanzo Dev](https://github.com/hanzoai/dev) agent, with this server attached |\n| `hanzo engine serve <model>` | local inference ([Hanzo Engine](https://github.com/hanzoai/engine)) |\n| `hanzo` | a coding session |\n\nThe native server takes no `serve` subcommand (`hanzo mcp serve` is tolerated).\n\n## Client Configuration\n\nNothing needs to be installed first: `npx -y hanzo mcp` fetches the launcher and the server.\n\n### 1. Claude Code\n\n```bash\nclaude mcp add --scope user hanzo -- npx -y hanzo mcp\nclaude mcp list\n```\n\n### 2. Codex\n\n```bash\ncodex mcp add hanzo -- npx -y hanzo mcp\n```\n\n### 3. Hanzo Dev (`hanzo dev`)\n\n[Hanzo Dev](https://github.com/hanzoai/dev) ([dev.chat](https://dev.chat)) attaches this server by default:\n\n```bash\nhanzo dev\n```\n\n### 4. Cursor, Windsurf, Zed, and VS Code\n\nAdd to your `mcpServers` configuration (`~/.cursor/mcp.json` or project `.mcp.json`):\n\n```json\n{\n  \"mcpServers\": {\n    \"hanzo\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"hanzo\", \"mcp\"]\n    }\n  }\n}\n```\n\nOnce you have run `hanzo auth login`, the server uses that sign-in; set `HANZO_API_KEY`\nin an `env` block only to override it.\n\n### 5. Over HTTP (SSE)\n\nFor remote or multi-agent deployments:\n\n```bash\nhanzo mcp -t sse --port 3333\n```\n\n### 6. Choosing tools\n\nEvery tool schema the server lists is sent to the model on each turn. Serve only the ones you use:\n\n```bash\nclaude mcp add --scope user hanzo -- npx -y hanzo mcp --tools fs,code,exec\n```\n\nRun `hanzo mcp --help` for every option (`--project-dir`, `--transport`, `--port`, `--tools`).\n\n---\n\n## Tool Catalog\n\nTool names and action schemas are the same in every runtime; [LLM.md](LLM.md) has the per-tool parity table:\n\n| Group | Tools | Description |\n| :--- | :--- | :--- |\n| **core** | `fs`, `exec`, `code`, `git`, `fetch`, `workspace`, `ui` | File read/write, shell execution, AST code edits, git operations, HTTP calls, UI components. |\n| **optional** | `think`, `memory`, `hanzo`, `plan`, `tasks`, `mode`, `gimp` | CoT scratchpads, semantic memory, Hanzo Cloud resource dispatch, task coordination, canvas. |\n| **code intelligence** | `code_search`, `code_context`, `code_ask`, `code_index` | Dense vector code search, symbol index lookup, context graph resolution. |\n| **tracker** | `tracker_boards`, `tracker_issues`, `tracker_create`, `tracker_update` | Visible engineering work items and live session status tracking on `hanzo.app`. |\n| **coding** | `coding` (`start`, `progress`, `changes`, `artifacts`, `tree`, `blob`, `steer`, `merge`, `stop`) | Autonomous coding runs: a repo and a task in, a pushed branch out; a pull request when the work is ready (Rust runtime). |\n| **kai (governance)** | `kai_decide`, `kai_choice`, `kai_score`, `kai_noul`, `kai_tool_select`, `kai_next_action`, `kai_command_guard`, `kai_injection_screen`, `kai_extract`, `kai_log_triage`, `kai_recover`, `kai_agent_step` | Typed policy evaluation, behavioral consistency benchmarking, command guards, failure recovery. The native server carries `kai_decide`; the rest are in the TypeScript server (below). |\n\n### The `hanzo` Cloud Dispatcher\n\nThe unified `hanzo` tool gives agents access to all 2,400+ endpoints across the Hanzo Cloud platform:\n- `resource`: Names the subsystem (e.g., `models`, `graph`, `iam`, `kms`, `billing`, `chain`, `network`).\n- `action`: Names the operation. Calling with no action lists available operations.\n- `args`: Arguments passed to the operation.\n\n```json\n{\n  \"resource\": \"models\",\n  \"action\": \"list\",\n  \"args\": {}\n}\n```\n\n### Kai Decision Primitives & Local Acceleration\n\nKai tools ask Hanzo's decision model typed questions (`POST /v1/decisions` on `https://api.hanzo.ai` or a local instance via `LOCAL_KAI_URL=http://127.0.0.1:8080`) and return calibrated probabilities and audit rationales:\nOnly `kai_decide` is in the native server; the others below are served by the TypeScript server.\n\n- `kai_choice`: Multi-class categorical decision matching policy criteria.\n- `kai_score`: Ordinal scoring and confidence assessment.\n- `kai_noul`: Evaluates the probability that a behavioral statement holds true.\n- `kai_command_guard`: Pre-execution command check (ALLOW, ASK, or DENY).\n- `kai_injection_screen`: Prompt injection and jailbreak barrier.\n- `kai_decide`: Batch evaluation of up to 100 questions in a single call.\n\nSee the [cookbook](cookbook/README.md) for 15+ runnable recipes.\n\n---\n\n## TypeScript server (`@hanzo/mcp`)\n\nThe TypeScript runtime is still published as [`@hanzo/mcp`](https://www.npmjs.com/package/@hanzo/mcp)\n(Node 18+) for anyone who wants it specifically, e.g. for the `kai_*` governance tools the\n[cookbook](cookbook/README.md) uses: `npx -y @hanzo/mcp serve`. Its own flags (`--core-only`,\n`--disable-tools <names>`, `--enable-ui`, `--enable-desktop`, `--full-surface`) are listed by\n`npx -y @hanzo/mcp serve --help`.\n\n---\n\n## Sibling Implementations\n\n- **Python**: [`hanzo-mcp`](https://pypi.org/project/hanzo-mcp/) on PyPI, built from [`hanzoai/python-sdk`](https://github.com/hanzoai/python-sdk).\n- **Rust**: the native server is [`rust/`](rust) in this repo (`hanzo mcp`); a native MCP client crate lives under [`hanzo-rs/sdk`](https://github.com/hanzo-rs/sdk).\n- **Go**: High-throughput MCP provider under [`hanzo-go/sdk`](https://github.com/hanzo-go/sdk).\n\n---\n\n## Development & Testing\n\n```bash\npnpm install\npnpm build\npnpm smoke    # runs smoke integration over stdio: initialize, tools/call, resources\n```\n\n---\n\n## Company & Leadership\n\nHanzo AI is co-led by Co-CEOs **Antje Worring** and **Zach Kelling**.\n\n- **Antje Worring**: Co-CEO ([LinkedIn](https://www.linkedin.com/in/antje-worring/) · [GitHub: zooqueen](https://github.com/zooqueen)) — corporate leadership, operations, ecosystem partnerships, and strategic execution.\n- **Zach Kelling**: Co-CEO & Founding CTO ([LinkedIn](https://www.linkedin.com/in/zeekay/) · [GitHub: zeekay](https://github.com/zeekay)) — foundational architecture, engineering, systems, and open-source models.\n- **Corporate Address**: 1 Ferry Building, San Francisco, CA 94105\n- **Inquiries**: [dev@hanzo.ai](mailto:dev@hanzo.ai) · [support@hanzo.ai](mailto:support@hanzo.ai)\n\n---\n\n## License\n\nMIT — see [LICENSE](LICENSE).\n"}