ActivityWatch MCP Server
ActivityWatch MCP server enabling AI agents to query ActivityWatch — the popular open-source, privacy-respecting time tracking and activity monitoring tool. Enables reading time-use data (which apps were used when, browser history, editor activity), analyzing productivity patterns, querying time spent by category, and integrating personal productivity data into AI-assisted reflection and planning workflows.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Local only. No external network. Personal activity data — highly sensitive. Privacy-first ActivityWatch keeps data on device. No credentials.
⚡ Reliability
Best When
A developer or knowledge worker using ActivityWatch for personal time tracking needs AI-assisted productivity analysis — getting insights from local activity data without sending it to cloud services.
Avoid When
You use a different time tracking system (Toggl, RescueTime, etc.) or need team-level rather than personal productivity data.
Use Cases
- • Querying how much time was spent on specific activities from productivity review agents
- • Analyzing daily and weekly productivity patterns from self-improvement agents
- • Getting context about recent work history for session summaries from assistant agents
- • Identifying time drains and focus patterns from coaching agents
- • Generating productivity reports from time-tracking data from reporting agents
- • Correlating work patterns with outcomes from personal analytics agents
Not For
- • Teams requiring centralized employee monitoring (ActivityWatch is for personal use)
- • Real-time activity monitoring for security (use SIEM/EDR tools for security monitoring)
- • Non-ActivityWatch time tracking systems (use Toggl, Clockify, or Harvest MCPs for those)
Interface
Authentication
No authentication — ActivityWatch runs locally and API is unauthenticated on localhost (port 5600). ActivityWatch must be running. Local-only access.
Pricing
ActivityWatch is free and open source. MCP server is free. Privacy-first: all data stays local.
Agent Metadata
Known Gotchas
- ⚠ ActivityWatch must be running for MCP to work — typically auto-starts with OS
- ⚠ Activity data includes sensitive personal information (websites visited, apps used) — handle carefully
- ⚠ AW's event bucketing system requires understanding bucket types (aw-watcher-window, aw-watcher-web, etc.)
- ⚠ Data accuracy depends on which watchers are running — configure required watchers first
- ⚠ Time zone handling is important — ActivityWatch stores in UTC; convert for local time queries
- ⚠ Large date range queries can return enormous amounts of data — implement time range limits
Alternatives
Full Evaluation Report
Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for ActivityWatch MCP Server.
Scores are editorial opinions as of 2026-03-06.