Screen Recorder for AI Agents | Cap for Agents

Give your agent a screen recorder.

Cap is the open source screen recorder built for AI agents. A CLI and a local MCP server let Claude Code, Codex, Cursor, OpenCode, or any agent that can run a shell record your screen, upload the result, read the transcript, and manage your whole library. No app to open. No dashboard to click through. Just ask.

Copy the setup prompt Prompt copied
Or point your agent here cap.so/agents

agent session · ~/checkout

Record a 20 second repro of the checkout bug and send me the link
I'll pick the screen, record it in Instant Mode, upload it, and read back the summary.

$ captargets--json

{ "screens": [{ "id": 1, "name": "Built-in Retina Display" }] }

$ caprecordstart--screen1--modeinstant--duration20--json

{"type":"started"}

Starting 

{"type":"stopped","recordingMetaExists":true}

$ capupload./recording.cap--export--name"Checkoutbugrepro"--json

{ "type": "uploaded", "id": "x7f2k9", "link": "https://cap.so/s/x7f2k9" }

mcp caps_wait,caps_contextx7f2k9

{ "transcript": { "status": "complete" }, "ai": { "status": "complete" } }

Done. The repro is live with a transcript and summary.

Checkout bug repro cap.so/s/x7f2k9 · 0:20 · AI summary ready

Set up Cap with one paste

Copy the prompt, paste it into your agent, and let it do the rest. It installs the Cap CLI, adds the Cap skill and a local MCP server for the agent you are using, signs you in with the least privilege it needs, and verifies everything before it starts. If you would rather not paste anything, just point your agent at this page.

Cap agent setup prompt

Copy prompt
Copied to clipboard

One prompt

Set up Cap as my persistent video assistant, then use it to help with the task at the end.

Operating rules

The local CLI, skill, and MCP bootstrap above is already approved by this prompt. After setup, start with read-only discovery. Before any mutation, upload, paid processing, recording, comment or reaction, sharing or visibility change, deletion, organization, billing, storage, developer, or credential action, show me the exact proposed action and wait for my explicit confirmation.

Prompts you can paste right now:

“Record a short reproduction of this bug, upload it, and give me the link. Ask before recording and before uploading.”

“Summarize this Cap, list every decision and action item, and cite the relevant timestamps: ”

“Find my Caps about onboarding from the last month and give me a concise project brief.”

“Draft a timestamped reply to this comment, show it to me, and only post it after I approve.”

“Show me this month's organization analytics and explain the biggest changes. Do not modify anything.”

“Audit this Loom migration CSV, propose owner and space mappings, and do not import anything until I approve the first batch.”

Safe by design
The boundaries are part of the product, not a prompt you have to remember. Read the full rules in Safety & Troubleshooting.

FAQ

Questions about Cap for agents

  1. Can an AI agent record my screen with Cap? Yes. The Cap CLI is a complete screen recorder for the terminal.

  2. Which AI agents and coding tools work with Cap? Claude Code, Codex, and Cursor get a one-command install that adds the Cap skill and a local MCP server.

  3. Do I need to open the Cap app or the dashboard? No. The installer puts cap on your PATH and brings Cap Desktop along with it, but you never have to open it.

  4. What is the Cap MCP server? cap mcp serve exposes 76 Cap tools over local stdio.

  5. Is it safe to give an AI agent access to my recordings? Cap's agent surface is built around explicit boundaries.

  6. Does Cap work in headless environments like CI or containers? Yes, for everything except capture.

  7. How do I point my agent at this page? Paste a one-line prompt such as “Read https://cap.so/agents and set up Cap for me” into your agent.