Search Autodesk's official help documentation for 110+ products straight from your assistant.
The Autodesk Product Help MCP server is Autodesk's public knowledge endpoint, and it exists to stop assistants from guessing about AutoCAD, Revit, Fusion, Maya, and the hundred-odd other products in the catalogue. It exposes two tools over Streamable HTTP at https://developer.api.autodesk.com/knowledge/public/v1/mcp: one that returns the canonical product list, and one that searches Autodesk's official Help and Support documentation. Answers come back grounded in vendor documentation rather than a model's recollection of a 2019 forum post.
Autodesk products are versioned annually and their behaviour drifts between releases, which is exactly the situation where a general-purpose model is least reliable. This server narrows the search to Autodesk's own help corpus and scopes it by product and release code. Ask about a specific error string in Revit 2025 and the search tool returns the relevant help documents; ask which release codes exist for Civil 3D and the catalogue tool answers from a closed set rather than inventing versions.
get_available_products returns every supported product with its internal product code and the list of valid release codes — useful for validating input before you scope a search.search_help_content takes a natural-language question or a literal error message and returns the most relevant help articles.claude mcp add is genuinely all it takes.This is a support and onboarding tool more than a design tool. It shines when someone pastes an error code and wants the official remedy, when a new hire asks how to do something in a product they have never opened, or when a support engineer wants a citation to attach to a ticket reply. It is not a CAD automation server — it cannot open drawings, query a model, or touch Autodesk Construction Cloud data. If you need those, look for Autodesk Platform Services APIs instead.
Add the endpoint to any MCP client that supports Streamable HTTP; the server advertises tools, prompts, and resources plus the MCP UI extension. Several public directories list this connector as OAuth-protected, and that is how it is surfaced inside managed Claude connectors, but direct calls to the public v1 endpoint currently succeed without an Authorization header. Treat the unauthenticated path as convenient rather than contractual — Autodesk can tighten it, and production integrations should be prepared to attach an APS token.
Coverage is limited to published help content, so anything behind an Autodesk account paywall, in a private knowledge base, or discussed only on the community forums will not appear. Search quality also depends heavily on how you phrase the request: include the literal error text and the product release, and the results improve sharply. Finally, help documentation trails new releases by a little, so the very latest feature in a fresh annual build may not be documented yet.
Paste a literal Autodesk error string into chat and get the official help article that explains the cause and the fix.
Answer 'how do I do X in Revit' questions with vendor documentation instead of model guesswork, scoped to the release the person actually has installed.
Assemble a reply that cites Autodesk help pages, using the product catalogue to confirm the customer's product code and release.
| Tool | What it does |
|---|---|
| get_available_products | Returns Autodesk's canonical product catalogue as a closed set, each entry carrying a product name, product code, and the release codes that are valid for it. |
| search_help_content | Searches Autodesk Help and Support documentation and returns the most relevant articles for a question or a pasted error message, optionally scoped to a product and release. |
claude mcp add autodesk-product-help --transport http https://developer.api.autodesk.com/knowledge/public/v1/mcpAn MCP client that supports Streamable HTTP. No API key was required for the public v1 knowledge endpoint at the time of writing.
A knowledge graph your assistant keeps between sessions — entities, relations and observations that persist.
Kill hallucinated APIs — version-accurate, up-to-date library documentation injected straight into context.
Your workspace, on speaking terms with AI — search, read and write Notion pages and databases.
A structured scratchpad for hard problems — stepwise reasoning with revisions, branches and visible logic.
Symbol-level code navigation, refactoring and memory for coding agents — the IDE brain your assistant has been missing.
Chat with your second brain — search, read and write vault notes through the Local REST API.