Back to browse
GitHub Repository
122 starsTypeScript

MemoryLane – Video as a token-efficient way to analyze user activity

by fidorka·Mar 3, 2026·3 points·1 comment

AI Analysis

●●●BangerBig BrainWizardry

Screenshots→summaries→MCP bridge gives AI assistants persistent context about your work.

Strengths
  • MCP integration is genuinely clever: embeds AI assistants directly into your development workflow without API proxying
  • Token-efficient approach: stores OCR and summaries locally, sends only one screenshot per summarization, reducing API costs
  • Privacy-respecting by design: screenshots deleted after processing, only metadata stored locally, users control API endpoint
Weaknesses
  • Requires paid OpenRouter subscription ($10/month managed key) or own API key; additional cost barrier
  • Windows/Mac only; no Linux support despite Linux being core dev environment for many developers
Target Audience

Software developers and knowledge workers using Claude, Gemini, or other MCP-compatible AI assistants

Similar To

Anthropic Claude Code · Zed AI · Continue IDE extension

Similar Projects

AI/ML●●Solid

I run a vision model on every screenshot, locally, on a 4GB GPU

Runs multimodal screen memory on 4GB VRAM while Microsoft Recall requires high-end NPU.

Dark HorseSolve My Problem
skye0110
3572d ago