Menú
Staleness detection and freshness-weighted retrieval: recency preferences by topic class, stale-source quarantine for pricing/policy/security, and re-ingestion signals. Owned by Metis.
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Basado en la clasificación ocupacional SOC
Offline A/B experimentation for prompt/policy/skill/rubric variants: defines what constitutes an experiment, the offline-only rule, the scoring contract, the anti-Goodhart + safety-floor constraints, the statistical-honesty rule, and the governance wiring into the TheEights evolution loop.
Encoded policy constraints: refunds/disclosures/consent, GDPR/CCPA data-minimization themes, right-to-human mandates, regulated-claim language rules, AI-disclosure requirements. Consumed by Harmonia, Soteria, Hermes, Eunomia.
The structured state token that travels with the customer across every agent handoff — identity ref, goal, active objects, constraints, sentiment, history digest. Minted by Iris; carried inside HANDOFF payloads.
Hard budgets, bounded loops, and the four terminal states. Step/dispatch/retry ceilings per command, graph-level termination conditions, and cost-tiering rules. Owned by Hestia.
The KPI frame: containment, FCR, AHT, CSAT-proxy, escalation precision/recall, cost-per-resolution, SLA attainment — and the Agent-Manager dashboard contract over events.jsonl. Owned by Hestia; consumed by Themis, Soteria, Echo.
Intent/language/sentiment/priority classification and routing rules for incoming support tickets. Owned by Iris; consumed by Hestia for dispatch.
| name | freshness-aware-retrieval |
| description | Staleness detection and freshness-weighted retrieval: recency preferences by topic class, stale-source quarantine for pricing/policy/security, and re-ingestion signals. Owned by Metis. |
| user-invocable | false |
| argument-hint | <check-freshness|weight-results> |
| allowed-tools | ["Read","Grep"] |
A stale source on the wrong topic is worse than no source: it produces confident, citable, wrong answers. This skill defines how the hearth weights recency and when age disqualifies a source outright.
Prevent stale-context failures (the research doc's "stale context and data drift" risk) by making freshness a first-class retrieval signal with topic-dependent thresholds.
kb-citation-grounding's source-freshness
dimension.| Class | Examples | Stale threshold | On stale |
|---|---|---|---|
| Volatile | pricing, plan limits, policy terms, security advisories, legal language | 90 days or any newer conflicting doc | source disqualified → treat as no source (Article VI.3) |
| Active | product features, UI walkthroughs, integrations, runbooks | 6 months | usable with an "as of " caveat in the draft |
| Stable | concepts, architecture, formats, troubleshooting fundamentals | 24 months | usable; prefer newer when scores tie |
When a topic spans classes, the most volatile component sets the class.
as-of-date — the date is part of the
citation form, so Themis and the customer can both see it.File a KB gap note (kb-gap-detection) tagged freshness when:
These notes are the change-data-capture signal the KB pipeline consumes; the squad does not edit the KB itself, it reports with evidence.