<< All versions

Skill v1.0.0

Trusted Publisher100/100
openai/community-plugins/analyze-log-signals
──Details
PublishedSeptember 16, 2026 at 10:51 AM
Content Hashsha256:7f763ba501bb89b5...
Git SHA
──Files
Files (1 file, 3.7 KB)
SKILL.md3.7 KBactive
SKILL.md · 45 lines · 3.7 KB

version: "1.0.0" name: analyze-log-signals description: Analyze reviewed Grafana log evidence for event patterns, error-rate changes, and enrolled security markers. Use when a user needs bounded log analytics for an incident window, error-log comparison, anomaly triage, operational debugging, or a cautious check of approved security signals.


Analyze Log Signals

Run the fixed grafana.logs pack and analyze only the normalized, redacted events and aggregates it returns.

Safety contract

  • Use only grafana_profile_status, grafana_list_packs, grafana_run_pack, grafana_get_evidence, and grafana_resolve_link.
  • Accept only profile and scope aliases; one declared LIVE, REPLAY, or SIMULATED origin; bounded current and baseline windows; and opaque references returned by these tools.
  • Never accept or construct LogQL, PromQL, SQL, regex searches, URLs, headers, datasource IDs, panel JSON, variables, or raw provider requests.
  • Do not request broader logs to compensate for an empty result. Scope expansion requires a separately reviewed server-side binding.
  • Keep logs, Grafana, alerts, datasources, and runtime state read-only. Do not reveal or reconstruct redacted content.
  • Confirm LIVE access and keep LIVE, REPLAY, and SIMULATED conclusions separate.

Workflow

  1. Establish profileAlias, scopeAlias, origin, and a current window no longer than seven days. Use an equal-duration, non-overlapping baseline when evaluating a change in error logs.
  2. Call grafana_profile_status({profileAlias}). Stop on ERROR or blocked readiness; disclose PARTIAL readiness.
  3. Call grafana_list_packs({profileAlias}) and require the exact active pack ID grafana.logs.
  4. Create one non-secret idempotency key and reuse it only for an exact retry.
  5. Call grafana_run_pack with that key, the aliases, packId: "grafana.logs", origin, current window, and optional baseline window.
  6. Review log event volume, log error rate, and any returned enrolled security-marker signal. Preserve event times, summaries, units, severity, direction, and ordering.
  7. Call grafana_get_evidence({reportRef,evidenceRef}) only for report-owned references needed to verify a pattern or issue. Summarize bounded patterns; do not reproduce sensitive event content unnecessarily.
  8. Call grafana_resolve_link({reportRef,linkRef}) only for a same-report link reference that helps a human inspect the enrolled Grafana view. Never accept a URL as input.

Use anomaly, regression, or security-event language only when a returned issue has classification: SIGNED_RULE and its returned kind and evidence support that label. A marker is not proof of exploitation, attribution, or impact. Treat classification: DEVIATION only as a deviation and state the uncertainty.

Interpret report status exactly

  • COMPLETE: expected log evidence completed; assess its returned signals rather than assuming there were no problems.
  • EMPTY: no usable events were returned for the approved scope and window; this is not proof that no events occurred.
  • STALE: events are too old for a current claim; report their observed and collected times.
  • PARTIAL: an expected signal, enrolled security marker, or baseline is missing, failed, or mixed; identify the gap.
  • CONTRADICTORY: evidence conflicts; cite both sides and do not collapse them into one narrative.
  • ERROR: collection failed or was rejected; report the sanitized error and stop.

Completion format

Return scope, origin, windows, overall status, event and error-rate summary with returned severity and direction, issues and security markers with their returned classification, gaps or contradictions, evidence references for every material claim, and only successfully resolved report-owned links.

All versionsv1.0.1 →