AI & developers

Use Frankreel from any AI assistant

Ask Claude, ChatGPT or Cursor to research keywords, score titles, write descriptions or check your channel, and it uses your Frankreel account to do it. One address, and you sign in once.

Your connector address

Paste this into any app that supports remote MCP servers. It asks you to sign in to Frankreel and approve.

https://frankreel.com/api/mcp

1 credit per AI tool, from your normal monthly allowance. Failed calls are never charged. Everything else is free.
Your data only. Nothing changes on YouTube unless you ask it to publish.
Disconnect any time in Settings, AI & API. Your Google sign-in is never shared.

Set it up

Claude (claude.ai, desktop and mobile)

Add Frankreel as a custom connector. You sign in to Frankreel and approve it; no key needed.

Open Customize, then Connectors (Team and Enterprise: an owner adds it under Organization settings, Connectors).
Choose Add custom connector.
Name it Frankreel and paste the server URL: https://frankreel.com/api/mcp
Click Add, then Connect. Sign in to Frankreel and press Allow.
In a chat, turn Frankreel on from the + menu, Connectors.

Server URL

https://frankreel.com/api/mcp

Claude Code

One command. Claude Code opens the browser to sign in the first time you use it.

Run the command, then type /mcp in Claude Code and choose Frankreel to sign in.

Terminal

claude mcp add --transport http frankreel https://frankreel.com/api/mcp

ChatGPT (apps and connectors)

Plans with developer mode can add Frankreel as an app over MCP, with OAuth sign-in.

Settings, Apps & Connectors, Advanced settings: turn on Developer mode (a workspace admin may need to allow it).
Back in Apps & Connectors, choose Create.
Name: Frankreel. MCP server URL: https://frankreel.com/api/mcp. Authentication: OAuth.
Create, sign in to Frankreel and press Allow.
In a chat, pick Frankreel from the + menu.

MCP server URL

https://frankreel.com/api/mcp

ChatGPT custom GPT (Actions)

For a custom GPT, import the OpenAPI spec and use an API key.

Create an API key in Frankreel: Settings, AI & API.
In the GPT editor: Configure, Create new action, Import from URL.
Paste: https://frankreel.com/api/v1/openapi.json
Authentication: API Key, Auth Type: Bearer, and paste your fr_live_ key.

OpenAPI URL

https://frankreel.com/api/v1/openapi.json

Cursor

Add to ~/.cursor/mcp.json (or .cursor/mcp.json in a project). Cursor signs you in with OAuth.

Save the file, then open Cursor Settings, MCP, and choose Login next to Frankreel.

mcp.json

{
  "mcpServers": {
    "frankreel": {
      "url": "https://frankreel.com/api/mcp"
    }
  }
}

VS Code (GitHub Copilot)

Add to .vscode/mcp.json, or run "MCP: Add Server" and choose HTTP.

Save the file and press Start above the server. VS Code asks you to sign in to Frankreel.

.vscode/mcp.json

{
  "servers": {
    "frankreel": {
      "type": "http",
      "url": "https://frankreel.com/api/mcp"
    }
  }
}

Gemini CLI

Add to ~/.gemini/settings.json. Uses an API key in a header.

Create an API key in Frankreel: Settings, AI & API.
Add the block below, then run /mcp in Gemini CLI to check the tools loaded.

~/.gemini/settings.json

{
  "mcpServers": {
    "frankreel": {
      "httpUrl": "https://frankreel.com/api/mcp",
      "headers": {
        "Authorization": "Bearer fr_live_YOUR_KEY"
      }
    }
  }
}

Other MCP clients and agents (n8n, Windsurf, Zed, Perplexity, your own)

Any client that supports remote MCP over Streamable HTTP works.

Server URL: https://frankreel.com/api/mcp
If the client supports OAuth sign-in, use it: discovery is automatic.
Otherwise send an API key: Authorization: Bearer fr_live_YOUR_KEY.
n8n: MCP Client Tool node, transport HTTP Streamable, Bearer auth with your key.

Server URL

https://frankreel.com/api/mcp

What it can do

The same 25 tools over MCP and REST. Every response says how many credits it used and how many are left.

Tool	What it does	Cost
`get_account` GET /me	Account, plan and credits	Free
`get_channel_overview` GET /channel	Channel stats and insights	Free
`channel_audit` GET /channel/audit	Channel audit	Free
`list_videos` GET /videos	List my videos	Free
`analyze_video` GET /videos/{video_id}	Analyze one of my videos	Free
`generate_titles` POST /videos/{video_id}/titles	AI title ideas for a video	1 AI credit
`generate_descriptions` POST /videos/{video_id}/descriptions	AI descriptions for a video	1 AI credit
`suggest_tags` POST /videos/{video_id}/tags	AI tag ideas for a video	1 AI credit
`save_draft` PATCH /videos/{video_id}/draft	Save a draft title, description or tags	Free
`publish_draft` POST /videos/{video_id}/publish	Publish a draft to YouTube	Free
`list_changes` GET /changes	Published changes and their results	Free
`keyword_research` GET /keywords	Keyword score	Free
`expand_keyword` POST /keywords/expand	Related keyword ideas (AI)	1 AI credit
`score_title` POST /titles/score	Score a title	Free
`score_thumbnail` POST /thumbnails/score	Score a thumbnail	Free
`find_outliers` GET /outliers	Find outlier videos	Free
`list_competitors` GET /competitors	Tracked competitors	Free
`competitor_videos` GET /competitors/videos	Competitor uploads	Free
`search_channels` GET /channels/search	Find a YouTube channel	Free
`add_competitor` POST /competitors	Track a competitor	Free
`remove_competitor` DELETE /competitors/{channel_id}	Stop tracking a competitor	Free
`list_ideas` GET /ideas	My video ideas	Free
`generate_ideas` POST /ideas	AI video ideas for a topic	1 AI credit
`create_blueprint` POST /ideas/{idea_id}/blueprint	Video blueprint for an idea	1 AI credit
`ask_coach` POST /coach	Ask the AI coach	1 coach message

REST API

Base URL https://frankreel.com/api/v1. Create a key in Settings, AI & API and send it as Authorization: Bearer fr_live_…. OAuth access tokens work too.

Success responses are { "result": …, "credits": { "charged", "remaining" } }. Errors are { "error": { "code", "message" } } with the HTTP status. Limit: 60 requests a minute per key.

OpenAPI 3.1: https://frankreel.com/api/v1/openapi.json. Import it as a ChatGPT GPT Action, a Gemini extension, or into Postman.

Your account, plan and credits left

curl https://frankreel.com/api/v1/me \
  -H "Authorization: Bearer $FRANKREEL_KEY"

Score a keyword (free)

curl "https://frankreel.com/api/v1/keywords?q=home%20workout" \
  -H "Authorization: Bearer $FRANKREEL_KEY"

Score a title (free)

curl https://frankreel.com/api/v1/titles/score \
  -H "Authorization: Bearer $FRANKREEL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"title": "I Tried Home Workouts for 30 Days", "keyword": "home workout"}'

AI title ideas for one of your videos (1 credit)

curl -X POST https://frankreel.com/api/v1/videos/VIDEO_ID/titles \
  -H "Authorization: Bearer $FRANKREEL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tone": "calm"}'

MCP: list tools

curl https://frankreel.com/api/mcp \
  -H "Authorization: Bearer $FRANKREEL_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Errors

Code	Status	Meaning
invalid_input	400	A parameter is missing or not valid. The message says which.
unauthorized / invalid_token	401	No key, or the key or token is wrong, revoked or expired.
plan_limit	402	Monthly allowance used up (or feature needs a paid plan). Nothing was charged.
insufficient_scope	403	A read-only key tried to act. Use a key with "Read and act" access.
needs_write_scope	403	Publishing needs YouTube publishing access, granted once in the web app.
not_found	404	No such video, idea or channel on your account.
no_channel	409	Connect a YouTube channel in Frankreel first.
rate_limited	429	60 requests a minute per key. Wait for Retry-After seconds.
youtube_quota / upstream_failed	503	YouTube or the AI service is unavailable (503 or 502). Nothing was charged; try again.

For builders

OAuth 2.1 with PKCE (S256), dynamic client registration and client ID metadata documents. Discovery: https://frankreel.com/.well-known/oauth-protected-resource/api/mcp and https://frankreel.com/.well-known/oauth-authorization-server.
Access tokens last 1 hour; refresh tokens 30 days and rotate on every use. Scopes: read, write.
The MCP server is stateless Streamable HTTP with JSON responses. Tools carry read-only and destructive hints.
For language models: /llms.txt and /llms-full.txt.