Agents

What it isCopy Anchor Link

Corpland Content runs an MCP server at https://content.api.corplandtechnologies.com/mcp. An agent connected to it can do what you do in the app: build a video with a Corpland playbook and read every beat of it, search the web, Pinterest and YouTube for pictures and footage, paint pictures, make clips, sound effects and music beds, read voiceovers in your cloned voice, transcribe audio with word timings, study a channel's style, write scripts, and read or save editor projects.

Every call spends the venture's credits exactly as the app does, at the prices on the billing page. Two actions exist only for agents: a transcription costs 2 credits a started minute, and a style study costs 10.

Agents is switched on per venture. If Agents is missing from Settings, ask us to turn it on.

Connect Claude.aiCopy Anchor Link

1. In Claude, open Settings, Connectors, Add custom connector.
2. Paste https://content.api.corplandtechnologies.com/mcp and connect.
3. A Corpland Content page opens. Sign in with your Corpland ID if you are not, pick the channel the agent should work in, and press Allow.

Claude is sent back connected. The connection shows under Settings, Agents in Corpland Content as an app, and Revoke ends it.

Connect Claude CodeCopy Anchor Link

Either sign in through the browser:

claude mcp add --transport http corpland-content https://content.api.corplandtechnologies.com/mcp

then type /mcp in Claude Code and choose Authenticate. Or make a key under Settings, Agents, New key and pass it as a header so nothing opens a browser:

claude mcp add --transport http corpland-content https://content.api.corplandtechnologies.com/mcp --header "Authorization: Bearer cc_key_..."

Connect Codex or any other clientCopy Anchor Link

Make a key under Settings, Agents. Clients that read an MCP config file want this shape:

{
  "mcpServers": {
    "corpland-content": {
      "type": "http",
      "url": "https://content.api.corplandtechnologies.com/mcp",
      "headers": { "Authorization": "Bearer cc_key_..." }
    }
  }
}

A key is shown once, when it is made. It works in the channel you had open when you made it and never expires until you revoke it.

Making a video from the chatCopy Anchor Link

Just ask for a video, or use the make a video prompt. The agent makes it itself and uses Corpland Content for the parts that need our services: a style card from your references, your voiceover, word timings, every picture, clip, footage slice and sound found or made through us with a stable URL. In Claude Code, Codex or any client that can run commands it then renders the cut on your machine with Remotion or ffmpeg. In Claude.ai or another client that cannot render, it writes the same timeline into the editor and gives you the link to finish and export in the browser. The timeline of any editor project is readable with get_edit, so an agent can also pick up a cut from the app and render it itself.

Using a Corpland playbookCopy Anchor Link

Ask for a playbook by name, or tell the agent to let Corpland do it, and it switches to the use a Corpland playbook prompt. It asks what the New video page asks: the playbook (Meme Commentary today), the voice (a voiceover made here or a recording on your machine), and the script it reads. Then it starts the build, tells you each stage as it happens, and gives you the video link when every beat has its picture.

What the agent cannot doCopy Anchor Link

Nothing renders an MP4 on our side. The editor exports in your browser, and a local agent renders on your machine. Thumbnails, publishing to YouTube and the Image Studio are not on the MCP yet.

SafetyCopy Anchor Link

• A connection acts as the person who made it, in one venture and one channel. When that person leaves the venture, every connection they made stops working.
• Keys and tokens are stored hashed. We cannot show a key again; make a new one.
• OAuth connections (Claude.ai, Claude Code signed in through the browser) expire after seven days and refresh on their own for ninety; revoking one in Settings ends it at once.
• Every tool that spends credits says so in its description, and the agent is told to warn you before it spends.