Agent Console
A local-first observability tool that reads Claude Code and Codex session transcripts to show token usage, cache performance, model choice, and cost across every machine a developer connects to a self-hosted team hub.
Agent Console is open-source, local-first observability for AI coding agents. It reads the Claude Code and Codex transcripts already stored on a machine and shows tokens, cache reads and writes, models, and list-price cost for every session, joining multiple machines, including laptops, build boxes, and servers, to a self-hosted team hub without double-counting anything.
It optionally ingests Claude Code OpenTelemetry data and gateway token metrics, exposes a Prometheus metrics endpoint and a Grafana dashboard, and clearly labels estimated figures as estimates and unknown readings as unknown rather than showing them as zero. A presenting mode replaces every project, machine, and person name with a stand-in for safe screen-sharing.
It runs from source with Git and a recent Node.js runtime, with no account, package install, or build step, opening a local browser console. Release builds are signed and notarized with published checksums, signed build attestations, and a software bill of materials.