vision-memory-mcp
MCP serverDocs & knowledgeCut your AI's vision token use by giving it a memory of UI screens it has already viewed. The app keeps that memory locally on your device and uses image fingerprinting and visual similarity to recognize repeats, so screens your AI sees again no longer need full re-analysis. The result is less visual processing on every task.
Unavailable. This server has no hosted endpoint yet, so ahel can't serve it.
After adding it, check the project repository at github.com/putervision/vision-memory-mcp for setup details. Then let your AI work on UI tasks as usual, and repeated screens will be served from its local cache instead of being processed from scratch.
What your AI can do with it
- Remember UI screens your AI has already viewed
- Recognize repeated screens using image fingerprints and skip re-analyzing them
- Match new visuals against saved ones using visual similarity
- Reduce vision token use across tasks that involve many UI images
- Keep the visual cache stored locally on your own device
Signals
- GitHub stars
- 62
- Forks
- 1
- Last commit
- Sep 2026
Advanced
- Delivery
- vision-memory-mcp MCP server → your ahel gateway (mcp.ahel.ai) → every connected AI client.
- Catalog kind
- mcp-server
- Gateway key
io-github-putervision-vision-memory-mcp- Source
- github.com/putervision/vision-memory-mcp