Open Chronicle brings local screen memory to Claude Code and Codex CLI
taoh released Open Chronicle, an open-source, Mac-only menubar app that captures and indexes screen activity locally via OCR and exposes it to coding agents like Claude Code and Codex CLI through an MCP server.
Score breakdown
Developers using MCP-compatible agents like Claude Code or Codex CLI can give their AI assistant persistent, fully local screen context — enabling richer, privacy-preserving agentic workflows without sending screen data to the cloud.
- 01Open-source macOS menubar app built as a local alternative to OpenAI Chronicle
- 02OCR powered by Apple Vision; summarization uses local AI providers via Vercel AI SDK — no data leaves the device
- 03Exposes an MCP server so any compatible coding agent can query screen history
Open Chronicle, released by taoh, is an open-source macOS menubar app designed to give coding agents like Claude Code and Codex CLI access to a local, persistent screen memory. The tool performs OCR using Apple Vision and runs summarization through local AI providers via the Vercel AI SDK, meaning no data leaves the user's machine. It is implemented in Swift for a low-footprint, efficient experience and exposes an MCP (Model Context Protocol) server so that any MCP-compatible coding agent can query the captured screen history.
taoh notes that the right balance between exclusion lists and allowlists is an open design question.
Privacy is addressed through a default blocklist that excludes password managers, messaging apps (Slack, WhatsApp, Messenger), and mail clients. taoh notes that the right balance between exclusion lists and allowlists is an open design question. On the limitations side, the tool is Mac-only by design, small local models with weak structured-output support will fail on `generateObject`, and retrieval currently uses LIKE-query keyword search. Full-text search (FTS) and optional embeddings are listed as planned improvements.
Key facts
- 01Open-source macOS menubar app built as a local alternative to OpenAI Chronicle
- 02OCR powered by Apple Vision; summarization uses local AI providers via Vercel AI SDK — no data leaves the device
- 03Exposes an MCP server so any compatible coding agent can query screen history
- 04Swift implementation designed for low memory and CPU footprint
- 05Default blocklist covers password managers, Slack, WhatsApp, Messenger, and mail clients
- 06Current retrieval uses LIKE-query keyword search; FTS and optional embeddings are planned
- 07Small local models with weak structured-output support will fail on `generateObject`