Zoo MCP

Zoo Model Context Protocol (MCP) Server

An MCP server housing various Zoo built utilities

Prerequisites

A Zoo account. Run uvx zoo-mcp auth login once before starting the server. The command opens your browser and prints a verification URL and code. The server reuses your saved login; public tools also work without login.

You can also use an API key:

export ZOO_API_TOKEN="your_api_key_here"

Installation

1. Ensure uv has been installed

1. Create a uv environment bash uv venv

1. Activate your uv environment (Optional)

1. Install the package from GitHub bash uv pip install git+ssh://git@github.com/KittyCAD/mcp.git

Running the Server

The server can be started by using uvx

uvx zoo-mcp

The server can be started locally by using uv and the zoo_mcp module

uv run -m zoo_mcp

Manage device login explicitly:

uvx zoo-mcp auth login

uvx zoo-mcp auth login --no-browser uvx zoo-mcp auth status uvx zoo-mcp auth logout

Logins are saved per API origin in your system credential store using keyring: macOS Keychain, Windows Credential Locker, or a supported Linux keyring. On headless systems without an available keyring, set ZOO_API_TOKEN instead. Status never displays tokens. Logout removes only the saved login for the current origin; restart running servers afterward. If your login expires or is revoked, run auth login again and restart the server. If the login command fails because of a network error, run it again to retry. Server startup never opens a browser. Device authorization and polling use requests-oauth2client. Device login requires the API's dedicated Zoo MCP device client registration.

The server can also be run with the mcp package

uv run mcp run src/zoo_mcp/server.py

Streamable HTTP

The server uses stdio by default. To serve the same tools over Streamable HTTP:

uvx zoo-mcp --transport streamable-http --host 127.0.0.1 --port 8000

From a local checkout:

uv run -m zoo_mcp --transport streamable-http

Connect an MCP client to http://127.0.0.1:8000/mcp. The SDK manages HTTP sessions and streaming responses. --host and --port configure the HTTP listener; their defaults are 127.0.0.1 and 8000.

Supplying client configuration from Python

Local entry points capture credentials when Zoo MCP is imported at startup. Token precedence is KITTYCAD_API_TOKEN, ZOO_API_TOKEN, then the registry's ZOO_TOKEN alias. Any explicit environment token bypasses saved/device login. API origin precedence is ZOO_HOST, KITTYCAD_HOST, then https://api.zoo.dev. Saved logins never cross API origins.

Applications can supply client configuration explicitly without changing the process environment or adding authentication arguments to MCP tools:

from zoo_mcp.client_config import ZooClientConfig, use_client_config

from zoo_mcp.server import mcp

with use_client_config(ZooClientConfig(token=api_token, base_url=api_origin)): result = await mcp.call_tool("list_org_datasets", {})

The context supplies the same token, origin, and headers to the KittyCAD SDK and KCL session constructors and restores the previous configuration on exit. New calls can use refreshed configuration. Finish child tasks before leaving their configuration context. An explicit configuration context also prevents the server lifespan from loading a saved local login. Hosted workers and embeddings should always use this interface.

To forward custom HTTP headers, pass headers={"User-Agent": "MyApp/1.0", "X-Request-Source": "example"} to ZooClientConfig. These headers reach REST requests and modeling WebSocket handshakes, including KCL sessions. A supplied User-Agent replaces the default value. Tokens and headers are excluded from configuration representations.

Client configuration applies to downstream calls. Applications serving different owners must also isolate their process-wide modeling sessions and workspaces.

Prebuilt binaries

Each GitHub release also attaches standalone executables (built with PyInstaller) for Linux (x86_64, arm64), macOS (arm64, x86_64), and Windows (x86_64) — no Python toolchain required. Download the binary for your platform, set ZOO_API_TOKEN, and run it directly, e.g.:

ZOO_API_TOKEN="your_api_key_here" ./zoo-mcp-linux-x86_64

The binaries are not code-signed, so macOS Gatekeeper and Windows SmartScreen may warn on first run.

Integrations

The server can be used as is by running the server or importing directly into your python code.

from zoo_mcp.server import mcp

mcp.run()

Individual tools can be used in your own python code as well. At Zoo we use zoo-mcp like this with ZooKeeper to save on resources. Instead of spinning up one MCP server per agent, each agent in a sense "embeds" the server in their own runtime. It has the additional benefit of preventing shared state.

from mcp.server.mcpserver import MCPServer

from zoo_mcp.zoo_tools import ResultZooExecuteKcl, zoo_execute_kcl

mcp = MCPServer(name="My Example Server")

@mcp.tool() async def my_execute_kcl(kcl_code: str) -> ResultZooExecuteKcl: """ Example tool that uses the zoo_execute_kcl function from zoo_mcp.zoo_tools """ return await zoo_execute_kcl(kcl_code=kcl_code)

The server can be integrated with Claude desktop using the following command

uv run mcp install src/zoo_mcp/server.py

The server can also be integrated with Claude Code using the following command

claude mcp add --scope project "Zoo-MCP" uv -- --directory "$PWD"/src/zoo_mcp run server.py

The server can also be tested using the MCP Inspector

uv run mcp dev src/zoo_mcp/server.py

For running with codex-cli

codex \

-c 'mcp_servers.zoo.command="uvx"' \ -c 'mcp_servers.zoo.args=["zoo-mcp"]' \ -c mcp_servers.zoo.env.ZOO_API_TOKEN="$ZOO_API_TOKEN"

You can also use the helper script included in this repo:

./codex-zoo.sh

The script prompts for a request, runs Codex with the Zoo MCP server, and saves a JSONL transcript (including token usage) to codex-run- .jsonl.

Architecture

Tools are defined in src/zoo_mcp/*.py, where they are then imported into src/zoo_mcp/server.py and tied to actual @mcp.tool() decorated functions.

src/zoo_mcp/zoo_tools.py acts as a large toolset to interact with Zoo's KCL and engine facilities. This source file houses other utilities like parse_unit or normalize_ext (normalizing file extensions).

Modeling scenes use explicit persistent sessions, with at most one session open per server process. Call get_modeling_sessions to recover its ID after a client reconnect, or call start_modeling_session when none exists. Populate the session with execute_kcl or import_cad_file, passing session_id to either tool. Reuse the same session_id with snapshot and modeling tools; then call stop_modeling_session when finished.

Sessions remain available across client reconnects while the server and engine connection stay alive. Idle sessions continue receiving engine status messages so keepalive checks remain responsive. Disconnected sessions are removed from get_modeling_sessions; start a new session and execute the model again to recover.

KCL execution and persistent modeling sessions use the CPU engine pool. Isolated KCL snapshots temporarily enable CPU rendering. Persistent sessions initialize rendering on the first snapshot, selection, highlight, or camera tool and retain it until the session is stopped, preserving visual state across calls. Exports, measurements, and geometry queries do not initialize graphics.

The KCL samples tools list and search public Aquarium projects from the Zoo API. Pass the returned name (a public project ID) to get_kcl_sample to download its KCL sources. Filenames preserve project-relative paths, including subdirectories. These public tools do not require an API token. The multipleFiles field in list/search results is a hint until the project has been downloaded.

execute_kcl runs mock execution before real execution and returns separate mock_preflight and real_execution objects. Each contains status (succeeded, failed, or not_run), message, and diagnostics grouped by severity. Stage messages are short summaries; the top-level message retains the full report for existing callers. Failed stages also expose error_family, including ZooMCPTimeoutError for session timeouts. Mock errors or an aborted mock execution return immediately with ok: false and real_execution.status: "not_run". Mock warnings remain in mock_preflight.diagnostics even if real execution fails. The known planeOf mock-engine limitation is reported as a warning so the real engine can evaluate it; other mock errors still block execution. Session responses expose mock diagnostics; the engine does not return real-stage diagnostics for session execution.

Path inputs capture the entrypoint, its transitive imports (including linked modules and glTF buffers), and project.toml once. Both stages use that copy without scanning unrelated files in the containing directory. Dependencies and symlink targets must stay inside the entrypoint's directory; external paths are rejected before file reads or execution. Transient local real-execution failures retain their bounded retries using the same copy without repeating mock execution. Diagnostics refer to the original source paths. Inline kcl_code accepts self-contained code and standard-library imports; filesystem imports require kcl_path so their dependencies can be captured within an explicit directory. When calling execute_kcl with session_id, check ok, then read path_artifact_graph on session success. The standalone mock_execute_kcl tool continues to return its existing boolean/message pair.

Contributing

Contributions are welcome! Please open an issue or submit a pull request on the GitHub repository

PRs will need to pass tests and linting before being merged.

ruff is used for linting and formatting.

uvx ruff check

uvx ruff format

ty is used for type checking.

uvx ty check

Testing

The server includes tests located in tests. To run the tests, use the following command:

uv run pytest -n auto