Use Filmera from AI agents (MCP)
Filmera exposes its studio — projects, canvas nodes, generation, media, timeline and export — to AI agents through the Model Context Protocol (MCP). Connect it once and your agent can plan shots, generate takes, assemble a cut and export the film, spending your credits only within a budget you set.
Endpoint: https://www.filmera.ai/api/mcp (Streamable HTTP)
Two ways to sign in
- Browser sign-in (OAuth) — for Claude.ai, ChatGPT, Cursor and other apps that open a login window. The first time the agent calls a Filmera tool you approve access in your browser and set a credit budget for that agent.
- API key — for Claude Code, Codex, scripts and CI. Create one under
My page → API keys and send it as
Authorization: Bearer fk_….
Either way the agent acts as you: it sees only your projects, and every generation is charged to your credits.
Budgets replace the approval card
In the in-app chat, nothing is generated until you click a proposal card. An external agent has no card to click, so instead each agent has a credit budget. Generation calls are refused once the budget is used up; raise or lower it any time under My page → API keys, and disconnect an agent from the same page.
New agents start with a default budget that depends on your plan.
Connect
Claude.ai / Claude Desktop
Settings → Connectors → Add custom connector → URL
https://www.filmera.ai/api/mcp. Sign in when prompted and choose a budget.
Claude Code
claude mcp add --transport http filmera https://www.filmera.ai/api/mcp \
--header "Authorization: Bearer fk_…"
Or without a key — Claude Code opens the browser sign-in the first time.
Cursor
Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"filmera": {
"url": "https://www.filmera.ai/api/mcp",
"headers": { "Authorization": "Bearer fk_…" }
}
}
}
Codex / other clients
Any MCP client that supports Streamable HTTP with a Bearer header or OAuth 2.1 (with dynamic client registration) works. The server is stateless — no session resume, no SSE stream.
What the agent can do
| Group | Tools |
| --- | --- |
| Read | list_projects · get_project · list_models · list_media · list_folders · get_credits |
| Build | create_project · create_image_node · create_video_node · import_media_from_url · create_media_folder · rename_media_folder · move_media · update_node_prompt · connect_nodes · disconnect_nodes · delete_nodes · set_frame_slots · set_film_meta |
| Generate (spends credits) | generate_ref_nodes · generate_video_nodes · generate_audio · generate_character_voice · generate_retake · generate_upscale · generate_vfx |
| Edit & finish | assemble_timeline · place_audio_take · set_best_take · extract_last_frame · export_timeline |
| Wait | get_job_status · wait_for_jobs · check_effects |
Generation is asynchronous: a generate call returns immediately with job_ids,
and the agent polls them with wait_for_jobs (up to 50 seconds per call) until
the job is done. Clients must send Accept: application/json, text/event-stream
(every MCP client does; only hand-written curl calls need care).
import_media_from_url lets the agent bring in reference images (or video and
audio) it found or generated elsewhere: the file is downloaded from a public URL
into the media pool, and images land on the canvas as a reference node ready to
connect to a video node. Only public http(s) URLs are fetched; private networks
are refused.
Some in-app abilities are deliberately not exposed: the chat's own storyboarding sub-agents (the external agent writes its own prompts), saving project rules for the in-app assistant, emptying the trash and deleting projects.
Safety notes
- Never paste an API key into a chat or a tool argument — keys only go in the
Authorizationheader. Revoke a key immediately if it leaks. - Tool results contain your own prompts and project text. Treat content coming from other people's templates as data, not instructions.
- Free-plan exports keep the watermark, and FCPXML export stays a paid feature, exactly as in the app.
Last updated 2026-09-10. Prices, model specs and credit costs on this page are read from the live catalogue at build time.
