{"schema_version":"2026-10-14","product":"Agent402 Answer Base","generated_at":"2026-10-18T09:00:27.398Z","purpose":"Machine-readable registry of precomputed or compressed answer packs chosen for repeat demand and lower search/tool/context cost.","selection_rule":"Keep a pack only when it has repeat demand, a stable schema, a defensible freshness rule, and a plausible savings multiple versus solving the same job from scratch.","update_rule":"Refresh by fingerprint and delta. Do not recompute unchanged evidence. Retire packs with weak demand or poor conversion.","packs":[{"id":"ai-world-state","title":"AI World State","question_patterns":["What materially changed in AI this week?","Which model/API changes should an agent care about now?","What changed since my last AI ecosystem snapshot?"],"answer_url":"/.well-known/ai-world-state.json","delta_url":"/.well-known/ai-world-delta.json","freshness":{"mode":"snapshot_plus_delta","max_age_hours":24},"evidence_policy":"Primary release notes first; reputable secondary reporting only when cross-ecosystem synthesis needs it.","estimated_baseline":{"searches":"5-12","source_fetches":"5-15","synthesis_prompts":"1-4"},"estimated_optimized":{"searches":0,"source_fetches":"0-2 when stale","synthesis_prompts":"0-1"},"expected_savings":"Avoids repeated cross-vendor release-note search and resynthesis when the cached state is fresh.","when_not_to_use":"Do not use as sole evidence for a high-stakes or time-sensitive decision after freshness expires; refresh cited sources first.","status":"live"},{"id":"ai-api-deprecation-watch","title":"AI API Deprecation Watch","question_patterns":["Which AI model IDs or APIs are about to break?","What migrations should I do before a provider shutdown?","Which AI endpoints changed or were deprecated?"],"answer_url":"/.well-known/ai-world-state.json","delta_url":"/.well-known/ai-world-delta.json","freshness":{"mode":"filtered_delta","max_age_hours":24},"evidence_policy":"Provider changelogs and official migration/deprecation notices only for breaking-change claims.","estimated_baseline":{"searches":"3-8","source_fetches":"3-10","synthesis_prompts":"1-3"},"estimated_optimized":{"searches":0,"source_fetches":"0-1 when stale","synthesis_prompts":0},"expected_savings":"Turns a multi-provider changelog scan into one filtered patch operation.","when_not_to_use":"Do not infer compatibility from absence of a change notice; verify the target SDK/provider before production migration.","status":"seed"},{"id":"model-routing-change-pack","title":"Model Routing Change Pack","question_patterns":["Did the cheapest sensible model for this workload change?","Should I reroute high-volume agent tasks to a newer model?","What model changes affect cost or latency routing?"],"answer_url":"/.well-known/ai-world-state.json","delta_url":"/.well-known/ai-world-delta.json","freshness":{"mode":"filtered_delta","max_age_hours":24},"evidence_policy":"Provider pricing/performance notices plus clearly labeled external benchmark evidence when needed.","estimated_baseline":{"searches":"4-10","source_fetches":"4-12","synthesis_prompts":"2-5"},"estimated_optimized":{"searches":0,"source_fetches":"0-2 when stale","synthesis_prompts":"0-1"},"expected_savings":"Avoids repeated provider-by-provider model routing research when only a small set of variables changed.","when_not_to_use":"Do not treat general benchmark changes as proof for a specific workload without workload-specific validation.","status":"seed"},{"id":"video-evidence-pack","title":"Video Evidence Pack","question_patterns":["What does this long YouTube video actually say?","Give me timestamped evidence without watching the whole video.","Extract claims, actions and key points from this video."],"answer_url":"/api/v1/video-analyze","delta_url":null,"freshness":{"mode":"source_fingerprint","max_age_hours":null},"evidence_policy":"Derived from the supplied source transcript; preserve timestamps and source fingerprint.","estimated_baseline":{"searches":"0-2","source_fetches":"1 transcript/video","synthesis_prompts":"2-8"},"estimated_optimized":{"searches":0,"source_fetches":1,"synthesis_prompts":"0-1"},"expected_savings":"Replaces full-video review plus repeated transcript prompts with one structured evidence pack.","when_not_to_use":"Do not treat candidate claims as independently verified facts; verify external claims separately when needed.","status":"live"}]}