Jam recordings (video, console, network) as context and proof of work for your coding agent.
Model Context Protocol (MCP) Server: dev.jam/jam
The dev.jam/jam MCP server provides jam recordings—video, console, network, and logs—as context for a coding agent. The server is described as capturing these recording types and making them available to support agent-driven development workflows.
🛠️ Key Features
Jam recordings as context
Includes video recordings
Includes console recordings
Includes network recordings
Includes log recordings
🚀 Use Cases
Provide coding agents with video-based execution context
Supply console output to support debugging and troubleshooting
Add network activity details to investigate integration behavior
A Jam is a screen recording or screenshot with the console logs, network requests and user actions attached, on one link. You'd record one for a bug report, a feedback walkthrough or a demo. These skills let your agent read a Jam, and record its own when it's done, so you check the result by watching it.
Works with Claude Code, Cursor, Codex, GitHub Copilot, Gemini CLI, OpenCode and every agent skills.sh supports. Add --skill jam-fix to install one skill.
Skills
You don't need to remember these names. Paste a Jam link, say what you want, and your agent picks the skill.
Give your agent a Jam
jam: "Take a look at this jam." For a link with no instructions. It reads the Jam and picks the skill below that fits.
jam-fix: "Fix this jam." Finds the root cause from the errors, failed requests and steps in the recording, and shows the evidence. It changes code only when you ask.
jam-to-tasks: "Do what I said in this jam." Turns your feedback into a numbered to-do list, each item tied to the moment you said it, then works through it.
jam-to-e2e: "Write a test for this jam." Writes an end-to-end test with your project's own test setup. A Jam of a bug becomes a test that fails until the fix lands.
jam-to-ticket: "Make a ticket for this jam." Files it in Linear, GitHub Issues or Jira with the steps, a frame from the recording and the errors, unless a ticket already exists.
video-to-jam: "Split this recording by bug." Turns an MP4, a Loom download or a long call into one Jam per issue.
Get a Jam back
jam-proof: "Record a jam showing it works." Records the change working and gives you a Jam link for the pull request.
jam-annotate: "Mark what's broken and share it." Draws boxes, arrows and notes on a screenshot, blurs anything private, and shares it as a Jam.
jam-proof-table: "Give me the jams for each fix." After a batch of work, one table: each claim, whether it works, and the Jam that shows it.
Set up Jam
jam-setup: "Set up jam." Installs the Jam CLI, signs you in, connects Jam MCP, installs these skills and runs jam doctor, in that order. It asks once whether to set up this project or every project on your machine.
Make a video
These need no Jam account. With the Jam CLI, a video can become a Jam link to comment on.
record-demo: "Record a demo of this feature." Plays the flow in a browser at human pace and renders a narrated 1440p video, with the camera moving only where the script says.
add-voiceover: "Add a voiceover to this video." Writes a script for the ear, voices it line by line and lays each line on the moment it describes, with a subtitle file.
record-terminal-proof: "Record proof that the migration ran." Records the commands and renders a GIF that plays inline on the pull request.
Requirements
A free Jam account, for every skill except the three video skills.
The Jam CLI. Say "set up jam" and jam-setup walks you through it. A skill signs in when it needs to. jam, jam-fix and jam-to-e2e can read a Jam through Jam MCP instead, the Model Context Protocol (MCP) server the plugin connects.
For video-to-jam: ffmpeg and a local transcriber such as Whisper.
For record-demo: agent-browser 0.38.1 or later, ffmpeg and Node 22 or later, with no npm packages. Its narration comes from add-voiceover.
For add-voiceover: ffmpeg and Node 22 or later, with no npm packages. An ElevenLabs key or the free Kokoro model (Python 3 and uv) gives a natural voice; without either, it uses the macOS voice.
For record-terminal-proof: asciinema 3 and agg. On Windows, use WSL.