Figma Context MCP
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
Descripción del creador (original en inglés)
Parametric 3D modeling in Autodesk Inventor: sketches, features, sheet metal, assemblies
Estos tres nunca se combinan en una sola puntuación: las opiniones, las mediciones y la popularidad son cosas distintas.
Todavía no escribimos un resumen sencillo para este. Abajo están los datos que leímos del registro.
Elige la app de IA que usas y copia la configuración.
Necesitas
¿Qué app usas?
Elige tu app arriba. La recordamos, así que la próxima vez el botón dirá “Instalar en …”.
Abre Claude Desktop → Settings → Developer → Edit Config y pega esto en claude_desktop_config.json
{
"mcpServers": {
"inventor": {
"command": "uvx",
"args": [
"inventor-mcp-server"
]
}
}
}Ejecuta esto en tu terminal
claude mcp add inventor -- uvx inventor-mcp-serverPégalo en .cursor/mcp.json de tu proyecto, o en ~/.cursor/mcp.json
{
"mcpServers": {
"inventor": {
"command": "uvx",
"args": [
"inventor-mcp-server"
]
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Pégalo en .vscode/mcp.json de tu proyecto
{
"servers": {
"inventor": {
"type": "stdio",
"command": "uvx",
"args": [
"inventor-mcp-server"
]
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Este se ejecuta en tu computadora, y ChatGPT solo puede conectarse a servidores en internet. Elige otra app.
Tu elección se guarda en una cookie en este dispositivo; no necesitas cuenta.
Esta configuración se genera a partir de la ficha del creador en el registro. Si el README dice otra cosa, sigue el README.
Se ejecuta en tu computadora; la lista de herramientas aparece en tu app de IA después de instalarlo.
Revisamos información que cualquiera puede ver; no hemos leído el código. Las revisiones que no aplican no se cuentan. · Revisado el 8 nov 2026
El nombre en el registro se verificó con la cuenta de GitHub cenitu, dueña del repo
Todavía no hemos leído este repo en GitHub
Todavía no hemos leído este repo en GitHub
Este se ejecuta en tu computadora; no hay endpoint en internet
Se ejecuta en tu computadora y usa ajustes locales en su lugar
Se ejecuta en tu computadora; solo podemos leer listas de herramientas de endpoints en internet
Comparado con todos los servidores activos de esta categoría en el registro. Las estrellas solo cuentan cuando quien lo publica es dueño del repo.
Superó 1 de 1 revisiones básicas aplicables. No hemos leído el código. Antes de agregar una clave o un token, revisa qué acceso pide el servidor y usa una clave que tenga solo los permisos que necesitas.
El registro no indica ninguna clave obligatoria, aunque algunos servidores te piden iniciar sesión la primera vez que los usas.
Se ejecuta en tu computadora. Tu app de IA lo inicia cuando hace falta, y los datos no pasan por los servidores del creador (a menos que el propio servidor llame a un servicio en línea).
El repo de GitHub cambió por última vez –, y la última versión (1.0.1) llegó al registro hace 7 días.
Las apps compatibles con MCP, como Claude Desktop, Claude Code, Cursor y VS Code. También necesitas Python (uv) en tu computadora.
En la misma categoría: Figma Context MCP, Claude Real Video, Figma MCP Server
Servidores de la misma categoría hechos por otros creadores, ordenados por estrellas.
Cómo leer una tarjeta:★ Estrellas de GitHubiTres pilares: reseñas · revisiones · usoiActualizado = fecha en que se publicó la última versión en el registroi
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
Let any LLM watch a video locally — and search everything it has ever watched.
The Figma MCP server brings Figma design context directly into your AI workflow.
Copiado tal cual del repo del creador. PopNx MCP no lo ha revisado ni lo respalda. Se quitaron HTML, imágenes y scripts. Los enlaces llevan fuera de este sitio.
Maintained continuation of [NeonGlay/inventor-mcp](https://github.com/NeonGlay/inventor-mcp).
The original repository has had no activity since its initial release (June 2026), and fresh
installs there fail on MCP SDK 2.x. This repo carries the full original history plus the fixes
and features that were submitted upstream as PRs and never reviewed. If the original author
returns, everything here is available to be merged back. See CHANGELOG.md.
MCP server for parametric 3D modeling in Autodesk Inventor — drive Inventor with Claude (or any MCP client) in natural language.
Build real parametric parts — flanges, shafts, nuts, brackets, sheet-metal enclosures — by talking to an AI assistant. The server wraps Inventor's COM API with high-level, millimeter-based tools and ships with battle-tested knowledge of Inventor 2026 API quirks that aren't documented anywhere else.
You: "Build a DIN 934 M16 hex nut with proper conical chamfers"
Claude: creates sketch → hexagon → extrude → tapped M16×2 hole → revolve-cut chamfers → done. Fully parametric, dimensioned sketches.
import_step (keeps the files in your project folder), placement with full orientation, check_interference / measure_distance, and sweep_rotation to check a door or lock moves freely through its whole rangeHole1 | V 23497 (-503 mm³) | F7 E14), so the AI can self-verify each stepfind_faces / find_edges work in a part and at any depth of an assembly, constrain takes origin geometry, faces and edges, and manage_* tools group what you do to documents, occurrences, features, parameters, constraints, materials and sheet metal rulessave_document saves only the active document unless asked, never a library or Content Center file; add your own paths with INVENTOR_MCP_PROTECTED or ~/.inventor-mcp/protected.txt. Deleting a component says what depends on it first, and reports what movedSilentOperation on a dedicated Inventor thread. A call stuck behind a modal dialog, or a long job, returns within 50 s with the dialog's title and text (read from the window list, no COM call); manage_call(wait_seconds=30) follows it — with the prints and progress() line so far — and returns the result the moment it finishes; manage_call(action="cancel") stops itexport_flat_pattern gives the flat size and writes the DXF/DWG for the laser (contour, holes and bend lines on their own layers); create_drawing makes a PDF drawing (base + projected + iso views, overall dimensions, or the flat pattern dimensioned); export_bom lists the assembly's parts with quantity, part number, description, material and mass (CSV for Excel / ERP); edit_properties reads and sets part number, description, stock number, material and custom propertiesdraw_closed_profile takes rounded corners ([x, y, r]) and arcs through a point, plus draw_rectangle(corner_radius), draw_slot, draw_arc and project_edges (exact edges of a body, for remodelling). Arcs go through three points, so a sketch with flipped axes can't mirror them; inspect_model warns when a curve was drawn as many short lines. draw_pattern draws grids and circles of holes / slots, offset_profile offsets loops, and check_sketch (also built into extrude's error) points at the gap that keeps a profile opencapture_view returns pictures of the model (iso / front / top / … , several in one call, zoom to components, isolate them) as images the agent can look at, with sketches hidden and the user's camera put back. A flange on the wrong side or a part floating off the frame shows at a glanceadd_frame_members lays Content Center profiles along skeleton lines (start → end, roll, centered or at the profile origin, shortened or extended at either end) and returns a saw cut list — what Frame Generator does, which has no API. A 13-member frame takes ~35 sconvert_surfaces_to_solid knits an imported surface model, patches the holes in the shell and closes it into a solid (or turns a closed shell into one directly), with the volume and any edges it couldn't closemanage_materials gives every matching part of an assembly a material (Content Center parts left alone), reports material and mass per part and flags Generic parts, surface-only vendor models and parts with no mass; edit_properties(mass_kg) overrides a massfind_faces finds faces of components inside sub-assemblies (type, normal, position, diameter) in top coordinates; constrain constrains them, corrects a wrong offset sign from where the face must land, and reports any other component the constraint dragged alonguse_content_center browses categories → families → sizes and makes members, custom length included (a 1500 mm DIN 1026-1 U 100 in one call); place_components places dozens of components with position, orientation, name and grounding in one callexecute_python errors show the traceback from the agent's own lines (also in helpers defined in earlier calls), stderr is captured, and undo_on_error=True rolls back everything the call did. load_library (or ~/.inventor-mcp/startup.py) keeps helper code loaded across namespace resets. COM errors come with their name and a hint (E_INVALIDARG, RPC_E_DISCONNECTED …) instead of a localized "Exception occurred". save_document checks the path first instead of failing with E_INVALIDARG and never rewrites library or Content Center files; manage_documents clears the Part12s left by failed attemptsconnect(project="…ipj") reuses the Inventor that has the project, or the running one if it is free, or starts a separate one; every tool takes instance= to pick one, and each has its own call thread and execute_python namespace, so one agent's long call doesn't hold up the other. list_instances shows them. An Inventor opened by hand can't be targeted — Windows always hands out the first one — so let the server start the second. A keeper process holds the Inventors the server started, so after a restart of the client connect(project=…) gets them backpip install "mcp[cli]" pywin32 — MCP Python SDK 1.x and 2.x are both supported (2.x renamed FastMCP to MCPServer; the server detects which is installed) git clone https://github.com/cenitu/inventor-mcp-server.git
pip install "mcp[cli]" pywin32
.mcp.json in your project root (see .mcp.json.example). Use the absolute path to inventor_mcp_server/server.py: {
"mcpServers": { "inventor": { "command": "python", "args": ["C:/path/to/inventor-mcp-server/inventor_mcp_server/server.py"] } } }
Don't use `"cwd"` with `-m inventor_mcp_server`. Claude Code ignores
cwdand starts the server from
your project root, where the package can't be found — the process exits immediately and the client only
reports
Connection closed. Pointing at the script by absolute path works from any directory.
Configs written for 0.2.x that point at
src/server.pystill work — that file now forwards to the
inventor_mcp_serverpackage — but please update them; the shim will be removed in a later release.
Alternatively, install it from PyPI — no clone needed, and the config gets a plain command:
pip install inventor-mcp-server
{ "mcpServers": { "inventor": { "command": "inventor-mcp-server" } } }
(If your MCP client can't find the command, use the full path printed by where inventor-mcp-server.)
Avoid `win32com.client.gencache.EnsureDispatch` and `CastTo` in your own scripts against the same Python install — they generate a
gen_pycache that breaks plainGetActiveObject(KeyError: '_dispobj_'). The server itself is immune since 0.5.0 (it never uses gen_py classes), but your scripts aren't: delete%LOCALAPPDATA%\Temp\gen_pyif it happens. See the API notes for the full story.
Claude Desktop reads the same mcpServers block from claude_desktop_config.json — use the full path of the Python that has mcp and pywin32 installed (a bare python may be a different one):
{
"mcpServers": { "inventor": { "command": "C:/Users/ /AppData/Local/Programs/Python/Python312/python.exe", "args": ["C:/path/to/inventor-mcp-server/inventor_mcp_server/server.py"] } } }
Two things that cost us an afternoon:
%APPDATA%\Claude\claude_desktop_config.json. It keeps its data in %LOCALAPPDATA%\Claude-Data\claude_ \claude_desktop_config.json (the is the package id, e.g. claude_pzs8sxrjxfjjc). If an edit seems ignored, look in %LOCALAPPDATA%\Claude\Logs\main.log — its paths show which folder the app really uses.Solo se muestra el principio; lee el resto en el repo