Marey MCP server

Marey

Give AI agents eyes for motion.

A screenshot tells an agent what the screen looks like. Marey tells it what changed.

Marey is a Model Context Protocol server that turns a short screen interaction into a timestamped, agent-readable contact sheet — a grid of still frames an agent can inspect as a temporal sequence, with any frame retrievable at full resolution.

claude mcp add marey -- npx -y @anilyesilkaya/marey

That's the whole install. Zero runtime dependencies, nothing to clone.

See it work

You: "The control point sometimes jumps when I drag it in the easing editor. Can you figure out what's happening?"

Agent: "Let me record the interaction with Marey." (records while you reproduce the drag)

Marey returns one image — the drag, decomposed into time:

Agent: "The cursor moves continuously, but watch the inspector's X value. Through #006 (1.25 s) the control point tracks the cursor exactly — both read x 331. At #007 (1.50 s) a snap guide appears at the 400 gridline while the point is still tracking at 343. One frame later, at #008 (1.75 s), the point snaps to X 400 px while the cursor has only reached 355 — a ~45 px discontinuity. From there the cursor keeps gliding and by #013 (3.00 s) has overtaken the point (415), yet the point stays pinned at 400. It isn't lagging — it's stuck to the line. Snapping is engaging too early; the snap threshold looks like it's being scaled by the 200% zoom, so it fires in canvas space instead of screen space. I'd check the snap-distance calculation."

A single screenshot would show a point and a cursor — and miss the jump entirely. The sequence makes the exact failing transition obvious, and the agent reads its diagnosis straight off the frames.

This is a real Marey contact sheet. Reproduce it with node demo/make-contactsheet.mjs, which renders the bundled fixture and composes the frames with Marey's own contact-sheet code.

Why Marey?

AI coding agents are good at understanding screenshots, but motion is still awkward. A bug such as:

• a control point jumping while it is dragged,
• a menu flashing and immediately closing,
• a canvas updating in the wrong order,
• an animation stuttering between states,

cannot be understood from a single screenshot. Marey bridges that gap: because it speaks MCP, the agent requests the recording itself and receives the resulting image directly in context.

No GIF inspection. No manually extracting frames. No dragging a dozen screenshots into chat.

It is named after Étienne-Jules Marey, a pioneer of chronophotography — the study of motion through sequences of images. Marey applies the same idea to AI agents: instead of handing a model a video it cannot reliably inspect frame by frame, it converts motion into a visual sequence the model can reason about.

How it works

1. An MCP client asks Marey to record the screen.
2. Marey captures frames at a chosen frame rate.
3. Each frame is numbered and timestamped.
4. Marey composes the frames into a contact sheet.
5. The contact sheet is returned to the agent as MCP image content.
6. Raw frames remain available on disk for closer inspection or re-stitching.

The idea is deliberately simple:

motion becomes one image containing time.

Tools

Marey exposes six MCP tools:

Shortened. The full README is on GitHub.