Bambu Lab MCP Server

Version: 1.0.0 · Follows Semantic Versioning

An MCP (Model Context Protocol) server that lets AI agents monitor and control Bambu Lab 3D printers. It is built on bambu-printer-manager and keeps one live MQTT session per printer. The same process serves a REST API and live camera streams. No Docker or container is needed.

Documentation: , with the full MCP tool reference and REST API reference.

---

Features

• MCP tools for discovery, state, print control, climate, filament and AMS, nozzles, detectors, files, camera, alerts and charts. Every tool carries a title and read-only, destructive, idempotent and open-world annotations.
• Write protection: every state-changing tool refuses unless called with user_permission=True, and tools that would disturb a print refuse while the printer is printing.
• A REST API over the same sessions, with a Swagger UI at /api/docs.
• Live camera streams for every model through the bambu-mse-stream relay: RTSPS H.264 passed through for X1, P2S and H2 series, TLS port 6000 stills encoded once for A1 and P1, played in the browser as fragmented MP4 (Media Source Extensions).
• Printer credentials encrypted at rest with AES-256-GCM, keychain-backed master key.
• 4 bambu://rules/* resources, a bambu://alerts/{name} alert feed and one prompt, bambu_system_context.

Security: the REST API and camera streams listen on all network interfaces with no authentication. Run bambu-mcp only on a trusted network.

---

Installation

Prerequisites

• Python 3.12+
• LAN access to your printers

Setup

git clone https://github.com/synman/bambu-mcp.git
cd bambu-mcp
python3 make.py

make.py creates .venv/ inside the project and installs the package with all dependencies. Run it again after any upstream dependency change.

Add your first printer

1. Discover: call discover_printers() to find printers on your network by SSDP.
2. Get access code: on the printer touchscreen, Settings, Network, Access Code.
3. Add: call add_printer(...) with the name, IP, serial and access code, and user_permission=True.
4. Verify: call get_printer_state(name="myprinter").

---

Client Configuration

Claude Desktop

Merge the "bambu-mcp" entry from config/claude_desktop.example.json into ~/Library/Application Support/Claude/claude_desktop_config.json, replacing ``.

GitHub Copilot CLI

Merge config/copilot_mcp.example.json into your Copilot MCP configuration.

Streamable HTTP

Run .venv/bin/python3 server.py --transport streamable-http and point the client at http://127.0.0.1:25099/mcp. Running it as a supervised daemon: docs/operators-guide.md.

---

Tools

104 tools (51 read-only, 53 write) and 87 REST routes. Full reference: bambu-mcp docs.

---

Architecture

bambu-mcp/
├── server.py            ← FastMCP entry point: tools, resources, prompt, startup
├── api_server.py        ← REST API and OpenAPI document
├── session_manager.py   ← Persistent BambuPrinter MQTT sessions
├── data_collector.py    ← Telemetry history per printer
├── notifications.py     ← State-change alert queue
├── secrets_store.py     ← AES-256-GCM secrets store
├── auth.py              ← Printer credential CRUD on top of secrets_store
├── job_project.py       ← Running job's resolved .3mf path per printer
├── port_pool.py         ← Shared port pool for the REST API and camera streams
├── daemon_port.py       ← Streamable HTTP port and singleton handling
├── gen_docs.py          ← Generates the published docs pages and the Tools section above
├── make.py              ← Cross-platform venv installer
├── camera/              ← camera relay, stream page, RTSPS and TCP-TLS clients, job monitor and analyzer
├── calibration/         ← Calibration scripts
├── tools/               ← Tool modules, plus _registry.py (registration and annotations)
├── resources/rules.py   ← bambu://rules/* readers (redacted)
├── prompts/context.py   ← bambu_system_context prompt
├── docs/                ← Operators guide, LAN file tunnel protocol, site overview source
└── config/              ← Example client configs and settings template

---

Versioning

The version of record is __version__ in server.py, plain semver with no suffix. pyproject.toml repeats it, and test_version.py fails when the two differ. The README header line and PLAN.md carry a copy that python make.py version-sync refreshes. The MCP initialize response reports it as serverInfo.version. The REST OpenAPI info.version and the mDNS record read the same constant. Nothing derives it from package metadata.

Raise it by hand with every change to a serving file:

• MAJOR: a tool removed or renamed, or a parameter or return shape changed incompatibly.
• MINOR: a tool or behaviour added.
• PATCH: a fix, hardening, or a wording change to a tool description, docstring or the server instructions.

Tests, docs, config, calibration/ and the standalone scripts (make.py, gen_docs.py, smoke_test.py, thermal_snapshot_monitor.py) need no bump. SERVING in hooks/version_bump_gate.py lists the files that do: the top-level modules the daemon imports and every .py under tools/, camera/, prompts/ and resources/. A pre-commit gate blocks a commit that stages one while __version__ is unchanged against HEAD. Install it once per clone: python3 hooks/version_bump_gate.py --install. The steps to release a change, and to read the version back from the running daemon, are in docs/operators-guide.md — Release a change.

---

Documentation

• Published docs: . The pages are generated by gen_docs.py from the tool docstrings, api_server.py and docs/site/overview.md. After changing any of those, run .venv/bin/python3 gen_docs.py and commit the regenerated pages in bambu-printer-manager.
• docs/operators-guide.md: daemon administration, restarts, logging and ports.
• docs/lan-file-tunnel-protocol.md: the port 6000 LAN file tunnel on H2-series, P2S and X2D.
• PLAN.md: implementation plan and design decisions.