AI Session Tracking
Query the session and run data Chau7 has recorded.
Availability
Partially implemented in stable. Read the limits below before relying on this workflow.
Reviewed against stable v0.5.1. Rust candidate additions are labeled separately. Release status.
The pain this solves
- Recorded agent activity can be difficult to query across terminal tabs.
- Session and usage history differs by CLI event and transcript support.
What ships with it
- MCP exposes current-session and stored-run queries.
- Run tools include lookups, lists, transcripts, and tool-call history.
- Details such as session IDs, transcript content, or usage metrics are present only when available from supported sources.
Questions this answers
- How to track AI coding sessions in my terminal?
- How long did my Claude Code session last?
- Session history for AI coding agent interactions
- Track AI agent runs and session duration in terminal
How it works
The stable app maintains terminal-session metadata and records telemetry runs from supported event sources. Available details can include provider, session ID, repository, timestamps, transcript or tool-call data, and usage fields when a source supplies them.
The MCP server exposes session_list and session_current along with run_get, run_list, run_transcript, and run_tool_calls. These queries let a connected client inspect available current sessions and stored run records.
Coverage is source-dependent. Detection of an AI command does not guarantee a complete transcript, provider usage, or lifecycle events for that CLI. OpenCode, for example, has a stable command/resume identity but no registry event-source value.
Why it matters
Session and run queries make recorded terminal-agent activity easier to inspect without implying that every CLI emits the same history. Chau7 exposes what its configured event and transcript sources provide.
Frequently asked questions
Can I list current sessions?
Yes. MCP provides session_list and session_current queries.
Does every detected agent have a full transcript and token history?
No. The available history depends on the CLI’s hooks, transcript format, proxy attribution, and other supported data sources.
Can I look up a past run?
Yes. MCP provides run lookup, list, transcript, and tool-call queries for stored telemetry runs.