Analytics
Analytics is the daemon’s current operational view. It is not a replacement for source data, workspace backups, or a monitoring system.
Operational endpoints
Section titled “Operational endpoints”These endpoints require analytics permission in authenticated deployments:
| Endpoint | Use |
|---|---|
GET /api/analytics/usage |
Request, actor, provider, and connector counters. |
GET /api/analytics/errors |
Recent errors; supports stage, time, and limit filters. |
GET /api/analytics/latency |
Latency summary. |
GET /api/analytics/logs |
Recent structured daemon log entries. |
GET /api/analytics/memory-safety |
Memory safety metrics. |
GET /api/analytics/continuity |
Continuity state. |
GET /api/analytics/continuity/latest |
Latest continuity summary. |
GET /api/telemetry/events |
Recorded telemetry event view. |
GET /api/telemetry/health |
Telemetry collector state. |
GET /api/timeline/* |
Timeline investigation routes. |
Start with a bounded query and correlate it with daemon status and diagnostics:
curl -fsS http://127.0.0.1:3850/api/analytics/errors?limit=20curl -fsS http://127.0.0.1:3850/api/analytics/latencycurl -fsS http://127.0.0.1:3850/api/diagnosticsCounters and in-memory buffers are runtime observations. They can reset when the daemon restarts. Use durable logs, database state, and the affected source when investigating a persistent incident.
Telemetry
Section titled “Telemetry”Daemon and CLI telemetry are configured at memory.pipelineV2.telemetryEnabled and default to off. Only explicit true enables collection. Missing means not yet answered; false means declined. Interactive setup, update install, and the dashboard ask once. You can keep it off explicitly with:
memory: pipelineV2: telemetryEnabled: falseUse signet telemetry to inspect consent, signet telemetry on or signet telemetry off to change it, or Settings → Advanced. Changes apply without restarting. Automated setup never prompts or writes a consent choice.
The npm package separately sends an anonymous install count before consent. Set SIGNET_TELEMETRY_OPTOUT=1 before installing to skip it. For a process-local opt-out that also suppresses consent prompts without editing configuration:
SIGNET_TELEMETRY_OPTOUT=1 signet daemon startThe daemon writes recorded telemetry events to:
$SIGNET_WORKSPACE/runtime/telemetry/events.jsonlPostHog delivery is best-effort and depends on telemetry configuration. The local JSONL audit is the inspection surface; it lets an operator see what was recorded without treating a remote dashboard as the source of truth.
Telemetry is designed to avoid memory content, user identity, file paths, and raw secrets. The audit log is still workspace-private operational data: do not commit it or paste it into a public issue without review.
Investigation workflow
Section titled “Investigation workflow”- Check daemon state and
/health/ready. - Pull a bounded recent error and latency view.
- Identify the affected request, session, source, provider, or queue.
- Inspect the original source and daemon logs.
- Use Diagnostics for health or repair decisions.
- Re-run the same bounded query after a change to verify the outcome.
Related: Daemon, Diagnostics, Configuration.

