speckle-mcp
This server connects AI agents to Speckle, enabling natural language management of BIM/CAD projects, models, and versions without transferring geometry data. It provides:
Project management: List all accessible projects with configurable limit (
speckle_projects_list), get project details including models and team (speckle_project_details), and create new projects with name, optional description, and visibility setting (speckle_project_create).Model management: List models within a project (
speckle_models_list), get model details including full version history with author, date, and commit messages (speckle_model_details), and create models categorized by discipline (e.g., Architecture, Structure) with optional description (speckle_model_create).Data transfer guidance: Obtain step-by-step, connector-specific instructions for sending data to Speckle from applications like Revit, Rhino, or AutoCAD (
speckle_send_instructions), and receive version info and guidance for pulling data into BIM apps or programmatic use (speckle_receive_instructions). These do not perform actual data transfer—they rely on native Speckle connectors.
Provides Speckle project and model management, with instructions for sending and receiving data via Archicad's Speckle connector.
Provides Speckle project and model management, with instructions for sending and receiving data via AutoCAD's Speckle connector.
Provides Speckle project and model management, with instructions for sending and receiving data via Blender's Speckle connector.
Provides Speckle project and model management, with instructions for sending and receiving data via QGIS's Speckle connector.
Provides Speckle project and model management, with instructions for sending and receiving data via SketchUp's Speckle connector.
Provides Speckle project and model management, with instructions for sending and receiving data via Unity's Speckle connector.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@speckle-mcp list all projects in my Speckle account"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
speckle-mcp
Connect Revit, Rhino, AutoCAD and more to AI agents. Manage projects, models and versions through natural language. Data flows through native Speckle connectors — this MCP provides instructions and metadata.
For Architects & Engineers
What does this do?
Instead of opening 4 different apps to check project status, ask Claude:
"Show me the latest model version in the Hospital project"
Claude connects to Speckle and tells you who sent what, when, and which discipline.
What can I ask?
You ask... | It does... |
"List my Speckle projects" | Shows all your projects with IDs and dates |
"What models are in project X?" | Lists architecture, structure, MEP models |
"Show me the version history" | Who sent what, when, with commit messages |
"Create a new project for the school" | Creates a project ready for your team |
"How do I send my Revit model?" | Step-by-step for Revit, Rhino, AutoCAD, Blender |
"How do I receive the latest model?" | Instructions for pulling data into your BIM tool |
How to install
Step 1: Install uv:
powershell -c "irm https://astral.sh/uv/install.ps1 | iex" # Windows
curl -LsSf https://astral.sh/uv/install.sh | sh # MacStep 2: Get a free Speckle token at https://app.speckle.systems/profile
Step 3: Download and configure:
git clone https://github.com/Daviidro/speckle-mcp.git
cd speckle-mcp
uv syncAdd to Claude Desktop config:
{
"mcpServers": {
"speckle": {
"command": "uv",
"args": ["run", "speckle-mcp"],
"env": { "SPECKLE_TOKEN": "your-token-here" }
}
}
}Is this safe?
Yes. Your Speckle data is protected by your personal access token.
You control which projects the token can access
Revoke the token anytime from your Speckle profile
The token stays on your computer — never sent anywhere except Speckle servers
Open source (MIT) — inspect every line
Related MCP server: mcp-server-for-revit
For Developers
Tools
Tool | Description |
| List all accessible Speckle projects |
| Get project details, models, and team |
| Create a new Speckle project |
| List models in a project |
| Get model details with version history |
| Create a new model with discipline |
| Guidance for sending from Revit/Rhino/etc. |
| Version info + guidance for receiving |
Supported platforms
Speckle Server: app.speckle.systems or self-hosted
Connectors: Revit, Rhino, Grasshopper, AutoCAD, Archicad, Blender, SketchUp, Unity, Unreal, Power BI, QGIS, Excel
Ecosystem
ifcopenshell-mcp — Talk to IFC models
bim-agent — Multi-MCP orchestrator
awesome-aec-mcp — Curated MCP list
License
Available Tools
8 toolsspeckle_model_createA
Create a new model within a Speckle project, organized by discipline (Architecture, Structure, etc.).
Inputs: name (required), project_id (required), discipline, description. Returns: result
speckle_model_detailsA
Get detailed model info including full version history with author, date, and commit messages.
Inputs: model_id (required), project_id (required). Returns: result
speckle_models_listA
List all models in a Speckle project. Models hold versioned BIM/CAD data from connectors.
Inputs: project_id (required), limit. Returns: result
speckle_project_createB
Create a new Speckle project to organize models shared across your team.
Inputs: name (required), visibility, description. Returns: result
speckle_project_detailsA
Get detailed information about a Speckle project, including its models and team members.
Inputs: project_id (required). Returns: result
speckle_projects_listB
List all accessible Speckle projects for the authenticated user. Projects are top-level containers that hold models.
Inputs: limit. Returns: result
speckle_receive_instructionsA
Get version info and instructions for receiving Speckle data into BIM applications or Python. Does NOT pull objects — use native connectors or specklepy operations.receive() for data retrieval.
Inputs: model_id (required), project_id (required), version_id. Returns: result
speckle_send_instructionsA
Get connector-specific instructions for sending BIM data to Speckle. Returns guidance for Revit, Rhino, AutoCAD, etc. Does NOT transfer data — use native Speckle connectors for actual data push.
Inputs: model_id (required), project_id (required), message. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 8 tools
Each tool targets a distinct resource (project vs model) and action (list/details/create), with send/receive instructions clearly separated. No two tools overlap in purpose; descriptions clarify any potential confusion between list and details.
All tools use a 'speckle_' prefix with resource and action, but pluralization is inconsistent (e.g., speckle_projects_list vs speckle_project_details) and the instruction tools use a verb-first pattern (speckle_send_instructions, speckle_receive_instructions). Overall pattern is readable and predictable.
8 tools is well-scoped for a server focused on project and model management plus data transfer instructions. Each tool serves a clear purpose without redundancy or excessive bloat.
The set provides create, list, and details for both projects and models, but lacks update and delete operations, leaving the lifecycle incomplete. Actual data transfer is not included, only instructions, which is a noted gap but perhaps intentional given the description.
Maintenance
Related MCP Connectors
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
MCP server that lets AI assistants use all OneSchema features exposed via the public API.
MCP server for building and testing AI agents with multi-model experimentation and insights.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceAI-powered MCP server that enables Claude and other LLMs to interact directly with construction documents, drawings, and specifications through advanced RAG and hybrid search capabilities.8MIT
- AlicenseNot gradedqualityDmaintenanceNode.js MCP server enabling AI assistants to interact with Autodesk Revit for model operations, data queries, and sketch-to-building generation via LLM.MIT
- FlicenseNot gradedqualityDmaintenanceEnables natural language control of Autodesk Revit through an AI-powered MCP server, allowing creation of walls, rooms, floor plans, and dimensions using conversational commands.-
- AlicenseAqualityDmaintenanceAn MCP server that lets users query, validate, convert, and compare IFC/BIM files using natural language. Runs fully locally and integrates with AI agents like Claude.8MIT