{"created_at":"2026-08-25T00:00:00.000Z","dossier_version":"0.1","manifest":{"format":"agentroute-promotion-dossier","payload_sha256":"sha256:3404c53c9b7990be7c238292fd967d2f65d03082fb509a445a30445603657683","root_sha256":"sha256:cfe6b2e003d0749fa6d2ddcd747585f61aff8a35e489c959fa66e4fb896c1532"},"payload":{"candidate_policy":{"criteria":{"max_cost_usd":0.05,"max_latency_ms":2000,"min_quality":0.8,"required_capabilities":["code"]},"id":"balanced-code-review","models":[{"model":"frontier-review-model","provider":"provider-a"},{"model":"fast-review-model","provider":"provider-b"}],"policy_version":"0.1","status":"reviewed","version":"1.0.0","weights":{"cost":0.2,"latency":0.2,"quality":0.6}},"compilations":[{"caveats":["Credential references and provider-specific deployment parameters must be added during human review.","The generated routing strategy is an approximation of multi-dimensional AgentRoute weights."],"compiler_version":"0.1","config":{"model_list":[{"litellm_params":{"model":"provider-a/frontier-review-model","order":1},"model_name":"balanced-code-review"},{"litellm_params":{"model":"provider-b/fast-review-model","order":2},"model_name":"balanced-code-review"}],"router_settings":{"routing_strategy":"simple-shuffle"}},"documentation":"https://docs.litellm.ai/docs/proxy/load_balancing","dry_run":true,"source":{"fingerprint":"sha256:e3fbbeda8f635f4815107de54394c95a87fc25cf14f60df46bc1dbe5bc2ef70d","policy_id":"balanced-code-review","policy_version":"1.0.0"},"target":"litellm"},{"caveats":[],"compiler_version":"0.1","config":{"policy":{"criteria":{"max_cost_usd":0.05,"max_latency_ms":2000,"min_quality":0.8,"required_capabilities":["code"]},"id":"balanced-code-review","models":[{"model":"frontier-review-model","provider":"provider-a"},{"model":"fast-review-model","provider":"provider-b"}],"policy_version":"0.1","status":"reviewed","version":"1.0.0","weights":{"cost":0.2,"latency":0.2,"quality":0.6}}},"documentation":"docs/evidence-suite-spec.md","dry_run":true,"source":{"fingerprint":"sha256:e3fbbeda8f635f4815107de54394c95a87fc25cf14f60df46bc1dbe5bc2ef70d","policy_id":"balanced-code-review","policy_version":"1.0.0"},"target":"native"},{"caveats":["OpenRouter provider preferences rank endpoints; latency preferences are not hard guarantees.","Review exact provider slugs before applying."],"compiler_version":"0.1","config":{"models":["provider-a/frontier-review-model","provider-b/fast-review-model"],"provider":{"allow_fallbacks":true,"order":["provider-a","provider-b"],"preferred_max_latency":2}},"documentation":"https://openrouter.ai/docs/guides/routing/provider-selection","dry_run":true,"source":{"fingerprint":"sha256:e3fbbeda8f635f4815107de54394c95a87fc25cf14f60df46bc1dbe5bc2ef70d","policy_id":"balanced-code-review","policy_version":"1.0.0"},"target":"openrouter"},{"caveats":["This is a review artifact, not a Portkey API payload.","Map provider names to workspace integrations before applying."],"compiler_version":"0.1","config":{"strategy":{"mode":"fallback","targets":[{"model":"frontier-review-model","provider":"provider-a"},{"model":"fast-review-model","provider":"provider-b"}]}},"documentation":"https://docs.portkey.ai/docs/product/ai-gateway/configs","dry_run":true,"source":{"fingerprint":"sha256:e3fbbeda8f635f4815107de54394c95a87fc25cf14f60df46bc1dbe5bc2ef70d","policy_id":"balanced-code-review","policy_version":"1.0.0"},"target":"portkey"},{"caveats":["Use the config inside an AI SDK generation call after human review.","Provider order applies independently to each fallback model."],"compiler_version":"0.1","config":{"model":"provider-a/frontier-review-model","providerOptions":{"gateway":{"models":["provider-b/fast-review-model"],"only":["provider-a","provider-b"],"order":["provider-a","provider-b"]}}},"documentation":"https://vercel.com/docs/ai-gateway/models-and-providers/provider-options","dry_run":true,"source":{"fingerprint":"sha256:e3fbbeda8f635f4815107de54394c95a87fc25cf14f60df46bc1dbe5bc2ef70d","policy_id":"balanced-code-review","policy_version":"1.0.0"},"target":"vercel-ai-gateway"}],"decision":{"analysis":{"arena_runs":1,"by_task_type":{"code_review":{"candidates":[{"candidate_id":"deep-review","mean_cost_usd":0.038,"mean_latency_ms":1416.666667,"mean_quality":0.926667,"model":"frontier-review-model","provider":"provider-a","samples":3,"success_rate":1,"successes":3},{"candidate_id":"fast-review","mean_cost_usd":0.009333,"mean_latency_ms":480,"mean_quality":0.91,"model":"fast-review-model","provider":"provider-b","samples":3,"success_rate":1,"successes":3}],"comparisons":[{"baseline_candidate_id":"deep-review","challenger_candidate_id":"fast-review","challenger_quality_losses":3,"challenger_quality_win_rate":0,"challenger_quality_win_rate_95ci":{"high":0.561497,"low":0},"challenger_quality_wins":0,"matched_pairs":3,"mean_cost_delta_usd":-0.028667,"mean_latency_delta_ms":-936.666667,"mean_quality_delta":-0.016667,"quality_pairs":3,"quality_ties":0}],"paired_units":3,"samples":6},"research":{"candidates":[{"candidate_id":"deep-review","mean_cost_usd":0.044667,"mean_latency_ms":1783.333333,"mean_quality":0.9,"model":"frontier-review-model","provider":"provider-a","samples":3,"success_rate":1,"successes":3},{"candidate_id":"fast-review","mean_cost_usd":0.015333,"mean_latency_ms":703.333333,"mean_quality":0.896667,"model":"fast-review-model","provider":"provider-b","samples":3,"success_rate":1,"successes":3}],"comparisons":[{"baseline_candidate_id":"deep-review","challenger_candidate_id":"fast-review","challenger_quality_losses":2,"challenger_quality_win_rate":0.333333,"challenger_quality_win_rate_95ci":{"high":0.79234,"low":0.061492},"challenger_quality_wins":1,"matched_pairs":3,"mean_cost_delta_usd":-0.029333,"mean_latency_delta_ms":-1080,"mean_quality_delta":-0.003333,"quality_pairs":3,"quality_ties":0}],"paired_units":3,"samples":6},"security":{"candidates":[{"candidate_id":"deep-review","mean_cost_usd":0.042667,"mean_latency_ms":1550,"mean_quality":0.946667,"model":"frontier-review-model","provider":"provider-a","samples":3,"success_rate":1,"successes":3},{"candidate_id":"fast-review","mean_cost_usd":0.013667,"mean_latency_ms":600,"mean_quality":0.933333,"model":"fast-review-model","provider":"provider-b","samples":3,"success_rate":1,"successes":3}],"comparisons":[{"baseline_candidate_id":"deep-review","challenger_candidate_id":"fast-review","challenger_quality_losses":3,"challenger_quality_win_rate":0,"challenger_quality_win_rate_95ci":{"high":0.561497,"low":0},"challenger_quality_wins":0,"matched_pairs":3,"mean_cost_delta_usd":-0.029,"mean_latency_delta_ms":-950,"mean_quality_delta":-0.013333,"quality_pairs":3,"quality_ties":0}],"paired_units":3,"samples":6},"summarization":{"candidates":[{"candidate_id":"deep-review","mean_cost_usd":0.020333,"mean_latency_ms":900,"mean_quality":0.893333,"model":"frontier-review-model","provider":"provider-a","samples":3,"success_rate":1,"successes":3},{"candidate_id":"fast-review","mean_cost_usd":0.005333,"mean_latency_ms":303.333333,"mean_quality":0.89,"model":"fast-review-model","provider":"provider-b","samples":3,"success_rate":1,"successes":3}],"comparisons":[{"baseline_candidate_id":"deep-review","challenger_candidate_id":"fast-review","challenger_quality_losses":1,"challenger_quality_win_rate":0.5,"challenger_quality_win_rate_95ci":{"high":0.905469,"low":0.094531},"challenger_quality_wins":1,"matched_pairs":3,"mean_cost_delta_usd":-0.015,"mean_latency_delta_ms":-596.666667,"mean_quality_delta":-0.003333,"quality_pairs":3,"quality_ties":1}],"paired_units":3,"samples":6}},"candidates":[{"candidate_id":"deep-review","mean_cost_usd":0.036417,"mean_latency_ms":1412.5,"mean_quality":0.916667,"model":"frontier-review-model","provider":"provider-a","samples":12,"success_rate":1,"successes":12},{"candidate_id":"fast-review","mean_cost_usd":0.010917,"mean_latency_ms":521.666667,"mean_quality":0.9075,"model":"fast-review-model","provider":"provider-b","samples":12,"success_rate":1,"successes":12}],"comparisons":[{"baseline_candidate_id":"deep-review","challenger_candidate_id":"fast-review","challenger_quality_losses":9,"challenger_quality_win_rate":0.181818,"challenger_quality_win_rate_95ci":{"high":0.476981,"low":0.051368},"challenger_quality_wins":2,"matched_pairs":12,"mean_cost_delta_usd":-0.0255,"mean_latency_delta_ms":-890.833333,"mean_quality_delta":-0.009167,"quality_pairs":12,"quality_ties":1}],"experiment_version":"0.1","generated_at":"2026-08-25T00:00:00.000Z","paired_units":12,"samples":24,"warnings":[]},"checks":[{"actual":12,"id":"global:matched_pairs","kind":"coverage","message":"global matched pairs: 12; minimum 12","metric":"matched_pairs","operator":"gte","scope":"global","status":"pass","threshold":12},{"actual":-0.009167,"id":"global:mean_quality_delta","kind":"threshold","message":"global challenger mean quality delta: -0.009167; minimum -0.02","metric":"mean_quality_delta","operator":"gte","scope":"global","status":"pass","threshold":-0.02},{"actual":0.051368,"id":"global:quality_win_rate_95ci_low","kind":"threshold","message":"global challenger quality win-rate 95% lower bound: 0.051368; minimum 0","metric":"quality_win_rate_95ci_low","operator":"gte","scope":"global","status":"pass","threshold":0},{"actual":-890.833333,"id":"global:mean_latency_delta_ms","kind":"threshold","message":"global challenger mean latency delta ms: -890.833333; maximum 0","metric":"mean_latency_delta_ms","operator":"lte","scope":"global","status":"pass","threshold":0},{"actual":-0.0255,"id":"global:mean_cost_delta_usd","kind":"threshold","message":"global challenger mean cost delta USD: -0.0255; maximum 0","metric":"mean_cost_delta_usd","operator":"lte","scope":"global","status":"pass","threshold":0},{"actual":0,"id":"global:success_rate_delta","kind":"threshold","message":"global challenger success-rate delta: 0; minimum 0","metric":"success_rate_delta","operator":"gte","scope":"global","status":"pass","threshold":0},{"actual":3,"id":"task_type:code_review:matched_pairs","kind":"coverage","message":"task_type:code_review matched pairs: 3; minimum 3","metric":"matched_pairs","operator":"gte","scope":"task_type:code_review","status":"pass","threshold":3},{"actual":-0.016667,"id":"task_type:code_review:mean_quality_delta","kind":"threshold","message":"task_type:code_review challenger mean quality delta: -0.016667; minimum -0.02","metric":"mean_quality_delta","operator":"gte","scope":"task_type:code_review","status":"pass","threshold":-0.02},{"actual":0,"id":"task_type:code_review:quality_win_rate_95ci_low","kind":"threshold","message":"task_type:code_review challenger quality win-rate 95% lower bound: 0; minimum 0","metric":"quality_win_rate_95ci_low","operator":"gte","scope":"task_type:code_review","status":"pass","threshold":0},{"actual":-936.666667,"id":"task_type:code_review:mean_latency_delta_ms","kind":"threshold","message":"task_type:code_review challenger mean latency delta ms: -936.666667; maximum 0","metric":"mean_latency_delta_ms","operator":"lte","scope":"task_type:code_review","status":"pass","threshold":0},{"actual":-0.028667,"id":"task_type:code_review:mean_cost_delta_usd","kind":"threshold","message":"task_type:code_review challenger mean cost delta USD: -0.028667; maximum 0","metric":"mean_cost_delta_usd","operator":"lte","scope":"task_type:code_review","status":"pass","threshold":0},{"actual":0,"id":"task_type:code_review:success_rate_delta","kind":"threshold","message":"task_type:code_review challenger success-rate delta: 0; minimum 0","metric":"success_rate_delta","operator":"gte","scope":"task_type:code_review","status":"pass","threshold":0},{"actual":3,"id":"task_type:research:matched_pairs","kind":"coverage","message":"task_type:research matched pairs: 3; minimum 3","metric":"matched_pairs","operator":"gte","scope":"task_type:research","status":"pass","threshold":3},{"actual":-0.003333,"id":"task_type:research:mean_quality_delta","kind":"threshold","message":"task_type:research challenger mean quality delta: -0.003333; minimum -0.02","metric":"mean_quality_delta","operator":"gte","scope":"task_type:research","status":"pass","threshold":-0.02},{"actual":0.061492,"id":"task_type:research:quality_win_rate_95ci_low","kind":"threshold","message":"task_type:research challenger quality win-rate 95% lower bound: 0.061492; minimum 0","metric":"quality_win_rate_95ci_low","operator":"gte","scope":"task_type:research","status":"pass","threshold":0},{"actual":-1080,"id":"task_type:research:mean_latency_delta_ms","kind":"threshold","message":"task_type:research challenger mean latency delta ms: -1080; maximum 0","metric":"mean_latency_delta_ms","operator":"lte","scope":"task_type:research","status":"pass","threshold":0},{"actual":-0.029333,"id":"task_type:research:mean_cost_delta_usd","kind":"threshold","message":"task_type:research challenger mean cost delta USD: -0.029333; maximum 0","metric":"mean_cost_delta_usd","operator":"lte","scope":"task_type:research","status":"pass","threshold":0},{"actual":0,"id":"task_type:research:success_rate_delta","kind":"threshold","message":"task_type:research challenger success-rate delta: 0; minimum 0","metric":"success_rate_delta","operator":"gte","scope":"task_type:research","status":"pass","threshold":0},{"actual":3,"id":"task_type:security:matched_pairs","kind":"coverage","message":"task_type:security matched pairs: 3; minimum 3","metric":"matched_pairs","operator":"gte","scope":"task_type:security","status":"pass","threshold":3},{"actual":-0.013333,"id":"task_type:security:mean_quality_delta","kind":"threshold","message":"task_type:security challenger mean quality delta: -0.013333; minimum -0.02","metric":"mean_quality_delta","operator":"gte","scope":"task_type:security","status":"pass","threshold":-0.02},{"actual":0,"id":"task_type:security:quality_win_rate_95ci_low","kind":"threshold","message":"task_type:security challenger quality win-rate 95% lower bound: 0; minimum 0","metric":"quality_win_rate_95ci_low","operator":"gte","scope":"task_type:security","status":"pass","threshold":0},{"actual":-950,"id":"task_type:security:mean_latency_delta_ms","kind":"threshold","message":"task_type:security challenger mean latency delta ms: -950; maximum 0","metric":"mean_latency_delta_ms","operator":"lte","scope":"task_type:security","status":"pass","threshold":0},{"actual":-0.029,"id":"task_type:security:mean_cost_delta_usd","kind":"threshold","message":"task_type:security challenger mean cost delta USD: -0.029; maximum 0","metric":"mean_cost_delta_usd","operator":"lte","scope":"task_type:security","status":"pass","threshold":0},{"actual":0,"id":"task_type:security:success_rate_delta","kind":"threshold","message":"task_type:security challenger success-rate delta: 0; minimum 0","metric":"success_rate_delta","operator":"gte","scope":"task_type:security","status":"pass","threshold":0},{"actual":3,"id":"task_type:summarization:matched_pairs","kind":"coverage","message":"task_type:summarization matched pairs: 3; minimum 3","metric":"matched_pairs","operator":"gte","scope":"task_type:summarization","status":"pass","threshold":3},{"actual":-0.003333,"id":"task_type:summarization:mean_quality_delta","kind":"threshold","message":"task_type:summarization challenger mean quality delta: -0.003333; minimum -0.02","metric":"mean_quality_delta","operator":"gte","scope":"task_type:summarization","status":"pass","threshold":-0.02},{"actual":0.094531,"id":"task_type:summarization:quality_win_rate_95ci_low","kind":"threshold","message":"task_type:summarization challenger quality win-rate 95% lower bound: 0.094531; minimum 0","metric":"quality_win_rate_95ci_low","operator":"gte","scope":"task_type:summarization","status":"pass","threshold":0},{"actual":-596.666667,"id":"task_type:summarization:mean_latency_delta_ms","kind":"threshold","message":"task_type:summarization challenger mean latency delta ms: -596.666667; maximum 0","metric":"mean_latency_delta_ms","operator":"lte","scope":"task_type:summarization","status":"pass","threshold":0},{"actual":-0.015,"id":"task_type:summarization:mean_cost_delta_usd","kind":"threshold","message":"task_type:summarization challenger mean cost delta USD: -0.015; maximum 0","metric":"mean_cost_delta_usd","operator":"lte","scope":"task_type:summarization","status":"pass","threshold":0},{"actual":0,"id":"task_type:summarization:success_rate_delta","kind":"threshold","message":"task_type:summarization challenger success-rate delta: 0; minimum 0","metric":"success_rate_delta","operator":"gte","scope":"task_type:summarization","status":"pass","threshold":0}],"decision_version":"0.1","evidence_sha256":"sha256:02e66478272850dfad98c9de61fb002dbf1b4ceb66e273b038f6c92dbf069aa5","generated_at":"2026-08-25T00:00:00.000Z","protocol_id":"public-proof-fast-review","protocol_sha256":"sha256:f2a05b181b5452247b0ead17b94b21d046dee842d4b3f34aac4a3fc18bdca5e6","status":"pass"},"gate":{"baseline_samples":12,"current_samples":12,"gate_version":"0.1","metrics":[{"current":12,"id":"baseline_samples","message":"baseline has 12 measured samples; 12 required","status":"pass","threshold":12},{"current":12,"id":"current_samples","message":"current has 12 measured samples; 12 required","status":"pass","threshold":12},{"current":1,"id":"observation_coverage","message":"current observation coverage 1; minimum 1","status":"pass","threshold":1},{"baseline":0.036417,"current":0.010917,"id":"mean_cost","message":"mean cost changed -70.022242%; maximum increase 0%","status":"pass","threshold":0},{"baseline":1412.5,"current":521.666667,"id":"mean_latency","message":"mean latency changed -63.067847%; maximum increase 0%","status":"pass","threshold":0},{"baseline":0.916667,"current":0.9075,"id":"mean_quality","message":"mean quality delta -0.009167; minimum -0.02","status":"pass","threshold":-0.02},{"current":0,"id":"policy_violations","message":"current policy violations 0; maximum 0","status":"pass","threshold":0},{"current":3,"id":"baseline_samples","message":"baseline has 3 measured samples; 3 required","slice":"task_type:code_review","status":"pass","threshold":3},{"current":3,"id":"current_samples","message":"current has 3 measured samples; 3 required","slice":"task_type:code_review","status":"pass","threshold":3},{"current":1,"id":"observation_coverage","message":"current observation coverage 1; minimum 1","slice":"task_type:code_review","status":"pass","threshold":1},{"baseline":0.038,"current":0.009333,"id":"mean_cost","message":"mean cost changed -75.439474%; maximum increase 0%","slice":"task_type:code_review","status":"pass","threshold":0},{"baseline":1416.666667,"current":480,"id":"mean_latency","message":"mean latency changed -66.117647%; maximum increase 0%","slice":"task_type:code_review","status":"pass","threshold":0},{"baseline":0.926667,"current":0.91,"id":"mean_quality","message":"mean quality delta -0.016667; minimum -0.02","slice":"task_type:code_review","status":"pass","threshold":-0.02},{"current":0,"id":"policy_violations","message":"current policy violations 0; maximum 0","slice":"task_type:code_review","status":"pass","threshold":0},{"current":3,"id":"baseline_samples","message":"baseline has 3 measured samples; 3 required","slice":"task_type:research","status":"pass","threshold":3},{"current":3,"id":"current_samples","message":"current has 3 measured samples; 3 required","slice":"task_type:research","status":"pass","threshold":3},{"current":1,"id":"observation_coverage","message":"current observation coverage 1; minimum 1","slice":"task_type:research","status":"pass","threshold":1},{"baseline":0.044667,"current":0.015333,"id":"mean_cost","message":"mean cost changed -65.672644%; maximum increase 0%","slice":"task_type:research","status":"pass","threshold":0},{"baseline":1783.333333,"current":703.333333,"id":"mean_latency","message":"mean latency changed -60.560748%; maximum increase 0%","slice":"task_type:research","status":"pass","threshold":0},{"baseline":0.9,"current":0.896667,"id":"mean_quality","message":"mean quality delta -0.003333; minimum -0.02","slice":"task_type:research","status":"pass","threshold":-0.02},{"current":0,"id":"policy_violations","message":"current policy violations 0; maximum 0","slice":"task_type:research","status":"pass","threshold":0},{"current":3,"id":"baseline_samples","message":"baseline has 3 measured samples; 3 required","slice":"task_type:security","status":"pass","threshold":3},{"current":3,"id":"current_samples","message":"current has 3 measured samples; 3 required","slice":"task_type:security","status":"pass","threshold":3},{"current":1,"id":"observation_coverage","message":"current observation coverage 1; minimum 1","slice":"task_type:security","status":"pass","threshold":1},{"baseline":0.042667,"current":0.013667,"id":"mean_cost","message":"mean cost changed -67.968219%; maximum increase 0%","slice":"task_type:security","status":"pass","threshold":0},{"baseline":1550,"current":600,"id":"mean_latency","message":"mean latency changed -61.290323%; maximum increase 0%","slice":"task_type:security","status":"pass","threshold":0},{"baseline":0.946667,"current":0.933333,"id":"mean_quality","message":"mean quality delta -0.013334; minimum -0.02","slice":"task_type:security","status":"pass","threshold":-0.02},{"current":0,"id":"policy_violations","message":"current policy violations 0; maximum 0","slice":"task_type:security","status":"pass","threshold":0},{"current":3,"id":"baseline_samples","message":"baseline has 3 measured samples; 3 required","slice":"task_type:summarization","status":"pass","threshold":3},{"current":3,"id":"current_samples","message":"current has 3 measured samples; 3 required","slice":"task_type:summarization","status":"pass","threshold":3},{"current":1,"id":"observation_coverage","message":"current observation coverage 1; minimum 1","slice":"task_type:summarization","status":"pass","threshold":1},{"baseline":0.020333,"current":0.005333,"id":"mean_cost","message":"mean cost changed -73.771701%; maximum increase 0%","slice":"task_type:summarization","status":"pass","threshold":0},{"baseline":900,"current":303.333333,"id":"mean_latency","message":"mean latency changed -66.296296%; maximum increase 0%","slice":"task_type:summarization","status":"pass","threshold":0},{"baseline":0.893333,"current":0.89,"id":"mean_quality","message":"mean quality delta -0.003333; minimum -0.02","slice":"task_type:summarization","status":"pass","threshold":-0.02},{"current":0,"id":"policy_violations","message":"current policy violations 0; maximum 0","slice":"task_type:summarization","status":"pass","threshold":0}],"slices":{"code_review":{"baseline_samples":3,"current_samples":3,"status":"pass"},"research":{"baseline_samples":3,"current_samples":3,"status":"pass"},"security":{"baseline_samples":3,"current_samples":3,"status":"pass"},"summarization":{"baseline_samples":3,"current_samples":3,"status":"pass"}},"status":"pass"},"promotion":{"reasons":["experiment protocol and routing quality gate passed with a reviewed candidate policy"],"verdict":"eligible"},"protocol":{"baseline_candidate_id":"deep-review","challenger_candidate_id":"fast-review","description":"Illustrative offline criteria for a faster review candidate across four task slices.","id":"public-proof-fast-review","minimum_matched_pairs":12,"minimum_slice_matched_pairs":3,"protocol_version":"0.1","quality_tie_tolerance":0.005,"required_task_types":["code_review","research","security","summarization"],"thresholds":{"maximum_mean_cost_delta_usd":0,"maximum_mean_latency_delta_ms":0,"minimum_mean_quality_delta":-0.02,"minimum_quality_win_rate_95ci_low":0,"minimum_success_rate_delta":0}}}}