LockedinLabs-AI/agent-console

The tool monitors local AI coding sessions to track token usage, cache states, selected models, and associated financial costs. It records data from the current machine and aggregates information from other devices linked through a self-hosted team hub. The system targets popular agents including Claude Code and Codex by providing detailed visibility into their operational metrics. This approach places data control directly with the user rather than relying on external cloud services for logging. Users install the application to begin capturing session data immediately on their local hardware. Additional machines connect to a central self-hosted hub to create a unified view of team-wide activity. The platform supports a presenting mode for reviewing aggregated statistics alongside policy hooks that manage specific operational constraints. Developments occur within a JavaScript codebase, and releases maintain MIT licensing standards for open access and modification. Prospective users should verify compatibility with Apple-signed and notarized macOS builds, which the publisher attributes to specific security signing processes. Release attestation details warrant review to confirm the integrity of downloaded binaries. While the project has logged 564 stars and 130 forks on GitHub, these figures provide limited context about long-term community support. The stated focus on local-first architecture remains an unverified claim until confirmed by independent technical audits.