Skip to main content

​
crewship metrics

Fetch mission performance metrics or pull a metric timeseries. Without flags, returns the mission summary — mission counts, completion time, and the token and cost totals of mission tasks in the trailing 24 hours. With --series <name>, switches to a timeseries endpoint and returns time-bucketed points.
These are MISSION metrics, and so are the money fields. total_cost_24h and total_tokens_24h sum mission tasks and nothing else. Spend outside a mission — chats, ad-hoc agent runs, routine runs — is not counted, so this command can report 0 on both while the workspace is genuinely spending. That is the design, not a broken query.A 0 here next to a non-zero crewship cost means no mission task was recorded in the window.
The Prometheus series are not here either. The crewshipd_* gauges, counters and percentiles — including the Issues & Routines §19.3 service levels for the mention → wake → reply loop (delivery latency, continuation, duplication, checkpoint compliance, outcome routing) and the crewshipd_issue_agent_sessions{state=…} gauge — are exposed on the server’s GET /metrics scrape endpoint for Prometheus, and there is no CLI command that reads them. See the Prometheus metrics reference.
Requires an authenticated session and a selected workspace (requireAuthAndWorkspace). The output is always structured JSON (or YAML with --format yaml). Pipe through jq or use --filter for ad-hoc shapes — there is no canonical table form for the records returned.

​
Flags

The global --format flag (table / json / yaml / ndjson / quiet) is honoured. The default is indented JSON because the data here is structured records without a single canonical “table” shape.

​
Endpoints

​
Examples

​
Mission summary

total_tokens_24h / total_cost_24h cover mission tasks only — see the warning above.

​
Active-runs timeseries over the last hour

​
Week-long window

​
Pull a single field via jq

​
YAML output

​
Pipe to a chart

​
See also

  • crewship cost — workspace-wide spend from the cost ledger. This is where the money is when total_cost_24h reads 0.
  • crewship paymaster — the same ledger rolled up per crew, agent, or mission.
  • crewship telemetry — distributed-trace observability across crews.
  • crewship journal — per-agent execution events behind the metrics aggregation.
  • crewship history — individual run records that feed the success-rate and duration percentiles.