agent-usage-analyze
v0.1.6
Published
Local-first usage, behavior, and improvement analytics for coding agents
Maintainers
Readme
Agent Usage Analyzer CLI
The CLI syncs local coding-agent sessions and starts the loopback-only dashboard. Codex has the most complete V1 task, evidence, automatic-capture, and advice path.
npx --yes agent-usage-analyze startThe command initializes private local storage, syncs supported Agent history, installs or refreshes Codex capture, registers the loopback dashboard as a macOS login service, and opens it. The command exits after the service is healthy, so the terminal can be closed. The first Codex history backfill continues in the background; the WebUI shows progress and only adds an ETA after throughput stabilizes. Use --wait-for-import to keep that work in the terminal. Trust the handler once from Codex /hooks if prompted. Use npx --yes agent-usage-analyze doctor for diagnostics and see the repository automatic analysis guide for advanced controls.
Product telemetry and remote semantic analysis are disabled by default. See the repository README, UPSTREAM, and LICENSE.
