ftrack MCP server
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
ftrack MCP server

A Model Context Protocol server that gives LLM agents (Claude Desktop, Claude Code, Cursor, …) broad, typed access to the ftrack Studio production-tracking API.
Why another one? The only existing ftrack MCP on GitHub is an early, unlicensed single-author experiment. This one is MIT-licensed, broad-coverage, and tested live against an ftrack Studio trial.
Coverage philosophy
ftrack's API is a generic query + CRUD over a flexible schema (143 entity types). So coverage comes from two layers:
Generic power tools —
query,query_one,create,update,delete— reach every entity type via ftrack's query language. This alone is ~full coverage.Typed convenience tools — projects, structure, tasks, statuses, assignments, notes, lists, time logs, thumbnails, users, and full schema introspection — make the common production ops one call each.
31 tools in total (see below).
Related MCP server: Devici MCP Server
Install
pip install -r requirements.txt # fastmcp, ftrack-python-api, requestsConfigure (credentials)
Set three env vars (or pass them in your MCP client config):
var | value |
|
|
| your login email |
| a Personal API key — ftrack ▸ avatar ▸ My account ▸ Security settings ▸ Create API key |
Run
python3 server.py # stdio transportWire into Claude Code
claude mcp add ftrack \
-e FTRACK_SERVER=https://yourstudio.ftrackapp.com \
-e FTRACK_API_USER=you@studio.com \
-e FTRACK_API_KEY=*** \
-- python3 /abs/path/to/ftrack-mcp/server.py(Tools appear as mcp__ftrack__* on the next session.) For Claude Desktop / Cursor, add the same command +
env to the app's mcpServers config.
Tools
Generic (full reach): query · query_one · create · update · delete
Every write takes
dry_run(defaultfalse).create/update/delete/set_statussupport two preview levels:dry_run="plan"(client-side echo, no server contact) anddry_run="preflight"— a real dry run that resolves every reference against live data, validates statuses against the schema, returns a before→after diff, and (oncreate) stages the op in ftrack's session to run its own schema validation, then rolls back — writing nothing. SetMCP_PLAN_LOG=/path.jsonlto capture a reviewable plan file. Preflight is high-confidence, not a guarantee (ftrack validates some rules only at commit). Other write tools takedry_runas a plain boolean. Schema / discovery:list_entity_types·get_entity_schema·list_project_schemas·list_statuses·list_task_types·list_object_types·list_priorities·list_custom_attributesProjects / structure:list_projects·get_project·create_project·list_children·list_tasks·create_taskOps:set_status·assign_task·add_note·get_notes·list_lists·log_timeMedia / versions:set_thumbnail·create_version·upload_review_media(encodes a movie into a web-reviewable AssetVersion) Cross-tracker:project_summary(normalized project snapshot — counts + per-shot status/thumbnail, canonical statuses — for verify/diff) Users / meta:whoami·list_users
Examples (what an agent would call)
"every in-progress lighting shot" →
query("Task where type.name is \"Lighting\" and status.name is \"In progress\"", ["name","parent.name"])"add a note to this shot" →
add_note("Shot", "<id>", "Looks good, ship it")"what fields does a Shot have?" →
get_entity_schema("Shot")"make a task" →
create_task("<shot_id>", "Compositing", status="Ready to start")
Part of a tracker-MCP quintet — migrate projects between platforms
This is one of five sibling tracker MCPs, each with the same shape (generic CRUD + schema + typed
convenience): shotgrid-mcp,
ftrack-mcp (this repo),
kitsu-mcp, and
ayon-mcp, and nim-mcp. They all speak the same production model
(Project → Sequence/Asset → Shot → Task → Version/Status), so an agent with two of them loaded can migrate
a project from one tracker to another — read the structure from the source MCP, recreate it via the
target's create/create_* tools, no bespoke script. This trio grew out of copying one project across all
three platforms.
📊 See COMPARISON.md for a side-by-side of the four trackers (data model, status
vocabularies) and the migration incompatibilities to know about — notably that ftrack has no
shot↔asset casting (so casting can't round-trip through ftrack) and its VFX schema lacks a clean
"done"/"approved" task status (lossy status mapping).
🧪 See TESTING.md for how these servers are validated — live round-trip tests, two-level
dry-run checks, and the cross-tracker migration matrix (including what is not yet covered, stated plainly).
Notes
Reads return entities serialized to the fields you request (dot-paths like
status.name), to avoid lazy-loading huge relations. Writes auto-resolve{"id": "..."}references (parent →Context, project →Project, status →Status, type →Type, …).Validated live: all 31 tools register (incl. video version media: encode_media -> ftrackreview-mp4); query/schema/create/update/delete/notes/status round-trips pass.
Writes are gated by
dry_run(preview, nothing staged). This replaced an earliercommitflag, wherecommit=falsewas a footgun — ftrack would stage the change and a later commit would flush it.A TS port over the official
ftrack-javascriptSDK is straightforward if you want it in-stack.
MIT © 2026 John Huikku
Built by John Huikku · alienrobot.com
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Every Canopy public API operation as a tool, run as you after a browser sign-in or on an API key.
Access Pollinations models and API capabilities through agent tools.
ADHD-friendly tasks, notes & projects for LucidNest - 18 tools, scoped tokens, Streamable HTTP.
Task and bug tracker for AI coding agents: 75 tools to file, triage and update work.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables LLMs to interact with Todoist task management platform through its API, supporting all features from the official Todoist TypeScript Client.MIT
- AlicenseCqualityDmaintenanceProvides LLM tools to interact with the Devici API, enabling management of threat modeling resources including users, collections, threat models, components, threats, mitigations, and teams.254MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI assistants to interact with Autodesk ShotGrid (Flow Production Tracking) for production tracking workflows. Provides 40+ tools for CRUD operations, batch processing, media, notes, and playlists.1,839 PyPI67MIT
- AlicenseNot gradedqualityDmaintenanceEnables LLMs to interact with Kitsu production tracker via its Zou API, providing tools for project management, asset and task operations, and cross-platform migration support.1MIT