Connect & Build · MCP + API

Build with
Autophocus.

One remote MCP server. Every studio capability as a tool call. Claude, ChatGPT, Codex — plug in and build with full creative control.

Quickstart →See the 11 tools

mcp-session

$ claude mcp add autophocus
transport: streamable-http
endpoint: api.autophocus.ai/mcp
connected · 11 tools
create_project
→ project_id: a83f…
generate_concepts
build_storyboard
write_scene
critique
export_project

OAuth 2.1 · auto

✓Discover auth server

✓Browser sign-in

✓Approve access

✓Token auto-refresh

Pipeline

Create

Ideate

Storyboard

Write

Refine

Export

create_projectok

{"name": "Neon Alley"}

→ project_id: a83f…

Export

screenplay.fountainscript

scene-data.jsonscenes

timeline.otioOTIO

final-cut.fcpxmlFCPXML

transportstreamable-http

tools11 available

authoauth 2.1

aphx_7Kp9…zQ

bearer token

How It Works

From prompt to final cut.

(01)Connect a real host

(02)Verify & authenticate

Prove the server is live, confirm tool discovery, then mint a personal token for MCP access.

(03)Create a project

Open a workspace and get back the project ID used by every later tool.

(04)Generate and refine

Chain ideation through drafting, then patch any scene without leaving the host.

(05)Export everything

Ship script, scene metadata, media files, OTIO, FCPXML, and EDL links.

Integration›Connect

ready

Connector — just the URL

claude mcp add --transport http

autophocus https://api.autophocus.ai/mcp

# no token — Claude opens sign-in ↗

Discover auth serveroauth 2.1

Approve accessoauth 2.1

Connected · auto-refresh11 tools

your session

> Add Autophocus to this workspace so I can create film projects from chat.
✓Connected to https://api.autophocus.ai/mcp
✓Autophocus tools are now available in the composer.
✓Next prompt can call create_project, generate_concepts, get_scenes, and export_project.

claude mcp add --transport http autophocus \
  https://api.autophocus.ai/mcp \
  --header "Authorization: Bearer aphx_xxx"

Integration›Health

checking

Phase 1

Verify the MCP server is healthy and list tools.

verified

MCP requests

healthz

GET /healthz

200 OK

oauth resource

GET /.well-known/oauth-protected-resource/mcp

issuer linked

tools/list

POST /mcp

11 tools

Live output

"status""ok"

"transport""streamable-http"

"tools"[11]

"resource""https://api.autophocus.ai/mcp"

Phase 2

Create a seven day token for MCP access.

Authenticated request

POST /api/v1/integrations/mcp/tokens/

Authorization: Bearer session_token

{ "expires_days": 7 }

waiting for response...

Live output

aphx_7Kp9...zQ

✓shown once

✓expires in 7 days

✓ready for header

HeaderAuthorization: Bearer aphx_7Kp9...zQcopy token

your session

> Verify the Autophocus MCP server is healthy and create a seven day token for local testing.
✓Health: ok
✓Transport: streamable-http
✓OAuth resource metadata resolved
✓tools/list returned 11 callable tools
✓Token created: aphx_7Kp9...zQ
✓Expires in 7 days
✓Use it as Authorization: Bearer aphx_7Kp9...zQ

curl -s http://127.0.0.1:9000/healthz | python3 -m json.tool
# {
#   "status": "ok",
#   "transport": "streamable-http",
#   "tools": [... 11 ...]
# }

MCP›create_project

project live

Tool call

create_project

name: "Neon Alley"

project_id: a83f...

Workspace created

01Conceptsready

02Storyboardready

03Screenplayready

04Scenesready

05Exportsready

your session

> Start a new project called Neon Alley about honest thieves planning a heist.
✓Tool called: create_project
✓Project created: Neon Alley
✓project_id: a83f...
✓Workspace sections ready: concepts, storyboard, screenplay, scenes, exports

res = await s.call_tool("create_project", {
    "name": "Neon Alley"
})
print(res.structuredContent or res.content)

MCP›generate + refine

11 tools

Active toolchain

Idea

Concepts

Storyboard

Screenplay

Scene patch

Critique

Live scene mutation

generate_concepts

generate

build_storyboard

generate

write_scene

generate

get_scenes

refine

update_scene

refine

critique

refine

generate_concepts4 angles

Noir comedy, contained heist, honest thieves

your session

> Generate concepts, turn the best one into a storyboard, write the opening scene, then make scene 1 tenser and funnier.
✓generate_concepts: 4 concept angles
✓build_storyboard: 9 scene cards
✓write_scene: 12 Fountain pages
✓get_scenes: 12 structured scenes
✓update_scene: scene 1 patched with tone and shot prompt

concepts = await s.call_tool("generate_concepts", {
    "project_id": project_id,
    "idea": "a heist where everyone is honest"
})
board = await s.call_tool("build_storyboard", {
    "project_id": project_id,
    "concept_artifact_id": concept_id
})
draft = await s.call_tool("write_scene", {
    "project_id": project_id,
    "storyboard_artifact_id": board_id
})

MCP›export_project

complete

export_project

project_id: a83f...

include media_files

include timeline_exports

screenplay.fountainscript

scene-data.jsonscenes

clip-001.mp4media

timeline.otioOTIO

final-cut.fcpxmlFCPXML

offline-edit.edlEDL

your session

> Export this project for handoff: script, scene data, media, and timeline files.
✓screenplay.fountain ready
✓scene-data.json ready
✓8 media files linked
✓timeline.otio, final-cut.fcpxml, and offline-edit.edl generated
✓Open in Autophocus URL included

exported = await s.call_tool("export_project", {
    "project_id": project_id
})
print(exported.structuredContent)

How It Works

From prompt to final cut.

(01)

Connect a real host

your session

> Add Autophocus to this workspace so I can create film projects from chat.
✓Connected to https://api.autophocus.ai/mcp
✓Autophocus tools are now available in the composer.
✓Next prompt can call create_project, generate_concepts, get_scenes, and export_project.

claude mcp add --transport http autophocus \
  https://api.autophocus.ai/mcp \
  --header "Authorization: Bearer aphx_xxx"

(02)

Verify & authenticate

Prove the server is live, confirm tool discovery, then mint a personal token for MCP access.

your session

> Verify the Autophocus MCP server is healthy and create a seven day token for local testing.
✓Health: ok
✓Transport: streamable-http
✓OAuth resource metadata resolved
✓tools/list returned 11 callable tools
✓Token created: aphx_7Kp9...zQ
✓Expires in 7 days
✓Use it as Authorization: Bearer aphx_7Kp9...zQ

curl -s http://127.0.0.1:9000/healthz | python3 -m json.tool
# {
#   "status": "ok",
#   "transport": "streamable-http",
#   "tools": [... 11 ...]
# }

(03)

Create a project

Open a workspace and get back the project ID used by every later tool.

your session

> Start a new project called Neon Alley about honest thieves planning a heist.
✓Tool called: create_project
✓Project created: Neon Alley
✓project_id: a83f...
✓Workspace sections ready: concepts, storyboard, screenplay, scenes, exports

res = await s.call_tool("create_project", {
    "name": "Neon Alley"
})
print(res.structuredContent or res.content)

(04)

Generate and refine

Chain ideation through drafting, then patch any scene without leaving the host.

your session

> Generate concepts, turn the best one into a storyboard, write the opening scene, then make scene 1 tenser and funnier.
✓generate_concepts: 4 concept angles
✓build_storyboard: 9 scene cards
✓write_scene: 12 Fountain pages
✓get_scenes: 12 structured scenes
✓update_scene: scene 1 patched with tone and shot prompt

concepts = await s.call_tool("generate_concepts", {
    "project_id": project_id,
    "idea": "a heist where everyone is honest"
})
board = await s.call_tool("build_storyboard", {
    "project_id": project_id,
    "concept_artifact_id": concept_id
})
draft = await s.call_tool("write_scene", {
    "project_id": project_id,
    "storyboard_artifact_id": board_id
})

(05)

Export everything

Ship script, scene metadata, media files, OTIO, FCPXML, and EDL links.

your session

> Export this project for handoff: script, scene data, media, and timeline files.
✓screenplay.fountain ready
✓scene-data.json ready
✓8 media files linked
✓timeline.otio, final-cut.fcpxml, and offline-edit.edl generated
✓Open in Autophocus URL included

exported = await s.call_tool("export_project", {
    "project_id": project_id
})
print(exported.structuredContent)

Setup

Authenticate in one step.

Recommended

OAuth 2.1 + PKCE

Paste the server URL. The host opens sign-in, you approve once, and the token auto-refreshes — nothing to copy or manage.

(01)Add api.autophocus.ai/mcp in the host
(02)Host opens the browser → you sign in
(03)Approve access on the consent screen
(04)Host receives the token via callback + auto-refreshes

See it in action →

Manual · custom hosts

Personal token

For scripts or hosts without OAuth support. Mint a token and pass it in the Authorization header.

terminal

# Authenticated as your Autophocus account:
curl -X POST https://api.autophocus.ai/api/v1/integrations/mcp/tokens/ \
  -H "Authorization: Bearer <session_token>" \
  -H "Content-Type: application/json" \
  -d '{"expires_days": 365}'

# → { "token": "aphx_…", "expires": "…" }   (shown once)

⚠️ Full account access — treat it like a password and revoke it if it leaks.

API Reference

Tools

11 MCP tools covering the screenwriter workflow. Filmmaker tools in development. Click any tool to see parameters, return types, and usage examples.

Write

Create and modify projects and scenes.

Read

Query projects, scenes, shots, and exports.

AI Generation

Generate concepts, storyboards, screenplays, and critiques.

Ready to connect?

Point any MCP host at https://api.autophocus.ai/mcp and go.

Book a Demo →

From prompt to final cut.

(01)Connect a real host

(02)Verify & authenticate

Prove the server is live, confirm tool discovery, then mint a personal token for MCP access.

(03)Create a project

Open a workspace and get back the project ID used by every later tool.

(04)Generate and refine

Chain ideation through drafting, then patch any scene without leaving the host.

(05)Export everything

Ship script, scene metadata, media files, OTIO, FCPXML, and EDL links.

> Add Autophocus to this workspace so I can create film projects from chat.

✓Connected to https://api.autophocus.ai/mcp

✓Autophocus tools are now available in the composer.

✓Next prompt can call create_project, generate_concepts, get_scenes, and export_project.

> Verify the Autophocus MCP server is healthy and create a seven day token for local testing.

✓Health: ok

✓Transport: streamable-http

✓OAuth resource metadata resolved

✓tools/list returned 11 callable tools

✓Token created: aphx_7Kp9...zQ

✓Expires in 7 days

✓Use it as Authorization: Bearer aphx_7Kp9...zQ

> Start a new project called Neon Alley about honest thieves planning a heist.

✓Tool called: create_project

✓Project created: Neon Alley

✓project_id: a83f...

✓Workspace sections ready: concepts, storyboard, screenplay, scenes, exports

> Generate concepts, turn the best one into a storyboard, write the opening scene, then make scene 1 tenser and funnier.

✓generate_concepts: 4 concept angles

✓build_storyboard: 9 scene cards

✓write_scene: 12 Fountain pages

✓get_scenes: 12 structured scenes

✓update_scene: scene 1 patched with tone and shot prompt

concepts = await s.call_tool("generate_concepts", { "project_id": project_id, "idea": "a heist where everyone is honest" }) board = await s.call_tool("build_storyboard", { "project_id": project_id, "concept_artifact_id": concept_id }) draft = await s.call_tool("write_scene", { "project_id": project_id, "storyboard_artifact_id": board_id })

> Export this project for handoff: script, scene data, media, and timeline files.

✓screenplay.fountain ready

✓scene-data.json ready

✓8 media files linked

✓timeline.otio, final-cut.fcpxml, and offline-edit.edl generated

✓Open in Autophocus URL included

> Add Autophocus to this workspace so I can create film projects from chat.

✓Connected to https://api.autophocus.ai/mcp

✓Autophocus tools are now available in the composer.

✓Next prompt can call create_project, generate_concepts, get_scenes, and export_project.

> Verify the Autophocus MCP server is healthy and create a seven day token for local testing.

✓Health: ok

✓Transport: streamable-http

✓OAuth resource metadata resolved

✓tools/list returned 11 callable tools

✓Token created: aphx_7Kp9...zQ

✓Expires in 7 days

✓Use it as Authorization: Bearer aphx_7Kp9...zQ

> Start a new project called Neon Alley about honest thieves planning a heist.

✓Tool called: create_project

✓Project created: Neon Alley

✓project_id: a83f...

✓Workspace sections ready: concepts, storyboard, screenplay, scenes, exports

> Generate concepts, turn the best one into a storyboard, write the opening scene, then make scene 1 tenser and funnier.

✓generate_concepts: 4 concept angles

✓build_storyboard: 9 scene cards

✓write_scene: 12 Fountain pages

✓get_scenes: 12 structured scenes

✓update_scene: scene 1 patched with tone and shot prompt

> Export this project for handoff: script, scene data, media, and timeline files.

✓screenplay.fountain ready

✓scene-data.json ready

✓8 media files linked

✓timeline.otio, final-cut.fcpxml, and offline-edit.edl generated

✓Open in Autophocus URL included

# Authenticated as your Autophocus account: curl -X POST https://api.autophocus.ai/api/v1/integrations/mcp/tokens/ \ -H "Authorization: Bearer <session_token>" \ -H "Content-Type: application/json" \ -d '{"expires_days": 365}' # → { "token": "aphx_…", "expires": "…" } (shown once)

Build withAutophocus.

From prompt to final cut.

From prompt to final cut.

Connect a real host

Verify & authenticate

Create a project

Generate and refine

Export everything

Authenticate in one step.

OAuth 2.1 + PKCE

Personal token

Tools

Write

Read

AI Generation

Tools

Write

Read

AI Generation

Ready to connect?

Build withAutophocus.

From prompt to final cut.

From prompt to final cut.

Connect a real host

Verify & authenticate

Create a project

Generate and refine

Export everything

Authenticate in one step.

OAuth 2.1 + PKCE

Personal token

Tools

Write

Read

AI Generation

Tools

Write

Read

AI Generation

Ready to connect?

Build with
Autophocus.

Build with
Autophocus.