io.github.kaiser-data/kitsune-mcp

---

The problem

When an agent writes an MCP server, it can't test its own change. The client still runs the old process, so after every edit a person has to restart the client or reconnect the server (and in some clients, register it in the config first) before the agent can call the tool and see whether the fix worked.

Kitsune hands that loop to the agent. connect, call and reload are tool calls, so the agent can run the tool, see it fail, fix the code, reload and check again on its own, in any MCP client.

Kitsune is one MCP server that sits in your client config and loads other servers while the session is running. Your own server under development, or any server from 7 registries. Their tools show up as native tools. Drop them and the session carries on.

connect("uvx --from . my-mcp-server", name="dev")  # start your server
shapeshift("dev")  # its tools appear in the client
call("summarize", arguments={"url": "https://example.com"})

# … the agent (or you) edits the tool …

reload("dev")  # restart on fresh code, remount
call("summarize", arguments={"url": "https://example.com"})  # new code, same session

Try it in 30 seconds with the planted-bug demo server: examples/demo-hot-reload/.

Install

Claude Code

claude mcp add -s user kitsune -- uvx kitsune-mcp

Any other client (Claude Desktop, Cursor, Windsurf, Cline, Zed, …):

{
  "mcpServers": {
    "kitsune": { "command": "uvx", "args": ["kitsune-mcp"] }
  }
}

Also on PyPI (pip install kitsune-mcp, then "command": "kitsune-mcp") and npm (npx kitsune-mcp, delegates to uvx).

Requirements: Python 3.12+ · uv · node/npx for npm-based servers · Docker optional, used to cage community servers.

Three things it does

	Loop	What you get
MCP REPL	connect → shapeshift → edit → reload → call	Iterate on your own server without restarting the client
Long-tail reach	search → shapeshift → call → shapeshift()	Use a server once without adding it to your config forever
Try before you trust	confirm=True + Docker cage + version pins	Run community npm/PyPI code contained, not as your user

Borrow a server you never configured

search("web scraping")
shapeshift("firecrawl", tools=["scrape_url"])  # mount one tool, not the whole server
call("scrape_url", arguments={"url": "https://example.com"})
shapeshift()  # drop it, session stays up

Try a community package, caged

search("pdf", registry="glama")
shapeshift("mcp-pdf-tools", confirm=True)  # npm/PyPI run in hardened Docker when available
call("extract_text", arguments={"path": "report.pdf"})
shapeshift()

Add credentials mid-session

auth("BRAVE_API_KEY", "sk-...")  # saved to ~/.kitsune/.env, no restart
shapeshift("brave", tools=["brave_web_search"])

Full live walkthrough: docs/demo-realtime.md.

When to use it, when not to

Use Kitsune when…	Skip it when…
You're building an MCP server and want an edit/reload loop	You only use 1–3 trusted servers (configure them natively)
A task needs a server that isn't in your config	Every turn hits the same server (keep it always-on)
You want to evaluate community MCP code without installing it for good	You want lower token use: the base cost is ~1,774 tokens/turn, and modern clients already defer tool schemas
Your MCP config is crowded (GATEWAY)	You need a sub-second first call (cold mount takes ~1–15 s)

How it works

shapeshift(server_id) picks a transport (stdio, HTTP+SSE, WebSocket, Docker), connects, reads tools/list, and registers each tool as a native tool with the server's real schema. The client receives notifications/tools/list_changed and sees first-class tools, with no wrapper in between. shapeshift() with no argument removes them and closes the connection.

search() covers the official servers repo, the MCP Registry, Glama, npm, PyPI, GitHub (github:owner/repo) and Smithery (hosted servers, free API key).

Tools

Default profile: 9 tools, ~1,774 tokens/turn (measured with python examples/benchmark.py).

Tool	Role
connect(command, name)	Start a persistent local server, e.g. the one you're developing
reload(name)	Release, restart on fresh code, remount, in one call
release(name)	Stop a persistent server
search(query, registry?)	Search 7 registries
shapeshift(server_id?, tools=[…])	Mount a server's tools (all or a subset); no argument drops them
call(tool_name, arguments)	Call a tool; the server is inferred when mounted
auth(var_or_server, value?)	Set an env key or run an OAuth 2.1 browser flow
auto(task, server_hint=)	Search, mount and call in one step
status()	Current form, connection pool, session stats

KITSUNE_TOOLS=all (or the kitsune-forge command) adds developer tools: inspect (schemas, credential check, token cost), test (quality score 0–100), bench (p50/p95 latency), compare, prewarm, craft, setup, and more. See docs/tools.md. Companion skills for Claude Code: kitsune-dev, kitsune-improve.

Safety, in short

• Community sources need consent. npm, PyPI and GitHub servers require confirm=True. The model can set that flag itself, so real approval belongs in your client's tool-approval UI.
• connect() runs the command it is given, as your user, with no consent gate and no cage. It is meant for your own code. Approve connect calls in your client before the agent runs them.
• Community packages run in a Docker cage by default when Docker is installed: no host filesystem, --cap-drop ALL, read-only root, memory and PID limits. Credentials are passed by name, never in argv.
• Version pins (TOFU). A later malicious publish doesn't silently replace a version you already ran.
• Opt-in action guard. A ~/.kitsune/policy.json checks every downstream call against allow/ask/deny rules, path roots and a host allowlist. ask goes to you through MCP elicitation.
• Install commands are validated and launched without a shell. fetch() blocks private hosts and re-checks every redirect.

Don't run it unattended with production admin, billing or security credentials. The full model, including what it does not protect against: docs/safety.md.

More

• Configuration: env and .env lookup, tool profiles, sandbox and trust settings, Smithery, GATEWAY, mount patterns
• Performance: cold/warm latency and token cost compared with always-on servers
• Transports · Compatibility · Agent patterns · Worked scenarios

Why "Kitsune"?

In Japanese folklore the kitsune (狐) is known for what it can become: it borrows a form, uses that power, casts it off and returns to itself. That is the product loop.

I am not Japanese, and I use this name with the highest respect for the mythology and culture it comes from. The parallel felt too precise to ignore.

Contributing

make dev     # install with dev dependencies
make test    # pytest
make lint    # ruff

Issues and PRs welcome: github.com/kaiser-data/kitsune-mcp · CHANGELOG.md

---

MIT License · Python 3.12+ · Built on FastMCP