Record a video
Turn a walkthrough into a demo or daily update. The agent checks readiness, records the useful steps, and waits for the finished video before returning a link.
lassu-record-videoGive your agent
a record button.
A demo of what you built. A screenshot that explains the bug.
Connect Lassu and let your agent show the work.
MacWindowsMCP + skills
Connect once. Then just ask.
Choose your AI tool. These three steps take you from setup to your first capture.
Get Lassu ready
Install the latest Lassu desktop app on the same Mac or Windows PC as your agent. Open it, sign in, and grant screen recording permission.
PreferencesAI capture & editingAllow AI accessOn Windows, open AI access in Preferences first. Turn on Allow AI access and keep Lassu running. The app includes its own runtime.
Connect Codex
In the same Lassu preferences, choose Connect Codex. This sets up the MCP connection and adds the recording and screenshot skills across your projects.
Connection + both skills included. No separate Codex CLI or Node installation needed.
Codex setup referenceManual setup on macOS
If you already use the Codex CLI and Lassu is in
/Applications, run this in Terminal. This registers MCP only; use Lassu’s Connect control to include the skills.Terminal · Codexcodex mcp add lassu -- "/Applications/Lassu.app/Contents/Resources/MCP/lassu" mcp serveTerminal · Check registrationcodex mcp listFor Windows or a different install location, use the Connect control in Lassu for the correct paths.
Start a new session. Check the connection.
Start a new local Codex session so it loads the connection and skills. Approve Lassu’s tools if your client asks, then send this message.
Check my Lassu connection, sign-in, screen permission, and recording allowance. Tell me which capture, editing, and narration features are available. Do not capture or record anything yet.
Your agent should report your connection, permissions, allowance, and supported features. Once those checks pass, try your first capture.
Lassu and your agent run on the same desktop. Your agent uses its own browser or computer tools to operate the app you want to capture.
The tools to capture. The skills to finish.
MCP connects your agent to Lassu. Skills teach it the workflow: choose a source, capture, edit, review, and share. Connecting Codex or Claude Code from Lassu includes both skills.
Edit a screenshot
Capture a window, add callouts, hide sensitive details, and review the result. Export it locally or upload it and choose who can see it.
lassu-edit-screenshotAdd narration when it helps
On supported Mac builds, your agent can trim pauses, arrange scenes, and add a voiceover in your chosen language. The skill guides it to review the rendered video and audio before uploading. Windows currently supports recording and screenshots; narration depends on the app’s reported capabilities.
Already using the Lassu plugin?
Keep it installed and connect through Lassu to set up MCP. If your client already manages an active Lassu plugin, Lassu uses those skills instead of installing a second copy. The plugin supplies instructions; the desktop app supplies the connection.
Lassu app updates refresh its managed skills. Plugin updates stay with your AI client, and customized files are preserved for review. Start a new agent session after updating.
Lassu skills on GitHubWhat do you want to show?
Copy a prompt and make it yours. Name the window, describe the result, and choose who can see it.
Make the screenshot explain itself
Use Lassu to capture my app window. Blur the email address, add an arrow and a numbered callout to the Save button, and put the image on a padded background. Show me the edited result and export a PNG locally.
Show what you shipped
Use Lassu to record my browser window while you demonstrate the new onboarding flow. Keep it under two minutes, then give me a link anyone with the link can watch.
Make a bug easy to see
Use Lassu to record my app window while you reproduce the search bug. Mark the step where it fails, then stop and upload the video. Keep it private.
Give your demo a voice
Use Lassu to make a short tutorial of my app’s onboarding flow. Check narration support first, record the useful steps, trim unnecessary waits, and add concise English narration. Review the finished video and audio before uploading, then give me a link anyone with the link can watch.
Explore the tools.
Ask in plain language; your agent handles the tools. Available features depend on your Lassu app and platform.
Access & sources2 tools
| Tool | What it does |
|---|---|
lassu_status | Check sign-in, screenshot access, screen permission, and recording allowance. |
lassu_list_sources | Find the display or app window to capture. |
Screenshots & editing11 tools
| Tool | What it does |
|---|---|
lassu_take_screenshot | Capture a display, window, or region as a local editable image. |
lassu_open_screenshot | Open a local image in a new editing session. |
lassu_list_screenshots | Find this agent’s local screenshot sessions. |
lassu_get_screenshot | Read the document and view its current image preview. |
lassu_edit_screenshot | Draw, redact, crop, resize, add backgrounds, compose images, and undo or redo. |
lassu_inspect_screenshot | Sample colors, measure pixels, and find smart selection bounds. |
lassu_screenshot_editor_action | Show or pin the editor, change the view or selection, and copy or paste. |
lassu_export_screenshot | Save PNG, JPEG, or a before/after GIF locally. |
lassu_upload_screenshot | Upload the reviewed image privately; get a job to track completion. |
lassu_get_uploaded_screenshot | Read an owned cloud screenshot’s current access and version. |
lassu_share_screenshot | Keep an uploaded image private or allow anyone with the link. |
Recording & sharing8 tools
| Tool | What it does |
|---|---|
lassu_start_recording | Start a private recording of a selected source. |
lassu_get_recording | Check capture progress and playback readiness. |
lassu_pause_recording | Pause the selected recording. |
lassu_resume_recording | Resume a paused recording. |
lassu_mark_recording | Save a labeled timestamp for a demo step. |
lassu_stop_recording | Stop, finalize, and upload the recording. |
lassu_retry_recording | Retry a stopped capture after a recoverable error. |
lassu_share_recording | Choose an audience and get the ready video's link. |
Narrated production7 tools
| Tool | What it does |
|---|---|
lassu_create_demo | Prepare a private narrated demo with a selected language and voice. |
lassu_update_demo | Arrange concise scenes and optional focus. |
lassu_finalize_demo | Render for review, then commit the inspected revision. |
lassu_get_demo | Inspect the current revision, media checks, preview and playback status. |
lassu_list_demos | Find existing demo drafts and their current state. |
lassu_inspect_demo_clip | Review a short video and audio clip from the exact rendered revision, when supported. |
lassu_discard_demo | Discard an unwanted, uncommitted demo draft. |
Narration & jobs5 tools
| Tool | What it does |
|---|---|
lassu_list_voices | Check narration availability, voices, and languages. |
lassu_prepare_narration | Generate speech segments before recording. |
lassu_attach_narration | Place prepared narration on the recording timeline. |
lassu_get_job | Check narration, video processing, or screenshot upload progress. |
lassu_cancel_job | Cancel speech generation or uncommitted demo work. Use stop to end a recording. |
You’re in control.
- Private until you share
- Agent recordings and uploaded screenshots start private. Ask to share with anyone who has the link, or with your workspace for videos. Returning a link does not send it to anyone.
- Your plan, same allowance
- Sign in to the Lassu desktop app to capture or edit. Video recordings use your existing allowance, including on Free. At the limit, Lassu prevents a new recording; screenshots and image editing remain available. Compare plans.
- Stop at any time
- Use Lassu’s visible recording controls to stop. Signing out or turning off Allow AI access revokes access and stops an active agent recording. MCP recordings also stop at their duration limit or after the initiating client loses contact for about a minute.
- Know what’s shared
- The connection runs locally. Screenshot previews go to your AI client; images upload to Lassu when requested. Finished videos upload for playback and sharing. Service-generated narration sends the script through Lassu’s speech service. Your Lassu login credentials stay out of the agent’s configuration. Privacy policy.
A little help getting connected.
Lassu is configured, but my agent can’t see it.
Keep Lassu running on the same desktop, sign in, and turn on Allow AI access. Start a fresh agent session to load the configuration and skills, then approve Lassu’s tools if prompted. If you moved or updated the app, reconnect from its preferences. “Configured” means the settings were written; the connection check in step three confirms your agent can use them.
Do I need a plugin, Node.js, or an API key?
The current Lassu app includes its own runtime. Connect Codex and Connect Claude Code set up MCP and both skills, so a separate plugin is optional. No global Node.js installation or API key is needed. If you already use the public Lassu plugin, the Connect flow keeps its skills managed by your client.
What if the Connect controls are missing?
Install a current Lassu build that includes AI access and its bundled runtime. In Preferences, look for AI capture & editing and Allow AI access. If the runtime is missing, reinstall the complete app. Copying a configuration cannot add features to an older build.
Can I use a web chat, Grok, or a remote agent?
Lassu needs a local client that can run a stdio MCP server on your Mac or Windows PC. Cloud-only chats, hosted connector URL fields, remote containers, and headless CI cannot connect directly. For a Grok model, use a local client that supports both that model and stdio MCP, then follow Other tools. The Gemini instructions are for Gemini CLI; Claude Desktop uses its local Developer configuration.
Why is my window or a feature unavailable?
Keep the target window open and ask your agent to refresh Lassu’s sources. On Mac, grant screen recording access in System Settings → Privacy & Security and reopen Lassu if macOS asks. Have the agent check status for the features supported by your build. Mac supports the full screenshot editor and optional narration; Windows exposes its supported recording and screenshot features.
Why is my video still processing?
Stopping a recording starts finalization and upload. Narrated demos also need rendering and review. The agent should wait for playable media before returning a link, and report any failed or interrupted job. If narration is unavailable, it should tell you before proceeding with a silent video.
How do I update or disconnect?
Update Lassu, then start a new agent session. If setup reports a conflict, close the AI client and retry; review any customized configuration or skills that Lassu preserved. To revoke access, turn off Allow AI access in Lassu preferences. You can also remove the Lassu MCP entry and optional plugin in your AI client.