{
  "schema": "cain.e128.workers.v1",
  "entity": "FIRME AI CORPORATION",
  "system": "CAIN-42",
  "evolution": "E128",
  "generated_at": "2026-11-21T17:20:55Z",
  "registered_identities_total": 100,
  "configured_specialized_workers": 8,
  "workers": [
    {
      "worker_id": "E128-W1-COGNITION",
      "registry_agent_id": "E127-001",
      "registry_group": "cognitive",
      "name": "COGNITION",
      "domain": "abstract_cognition",
      "target_trials": 250,
      "config_version": "e128.v1",
      "config_hash": "0fc776e9e99fe962e2b65906b612264e2cf3ad480bc12b141e3b9b7dc351c451",
      "model_provider": "ollama",
      "model_id": "qwen2.5:0.5b",
      "model_version": "ollama:0.12.3:494.03M:Q4_K_M:a8b0c5157701",
      "model_available": true,
      "system_prompt_hash": "21f5f85689c7dc903c066ed67b471e1fd84ba930ad795ae27904f1a91082d0d0",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": false
      },
      "research_specialization": [
        "abstract_reasoning",
        "compositional_reasoning",
        "rule_induction",
        "concept_acquisition",
        "novel_problem_solving",
        "multi_step_reasoning",
        "metacognition"
      ],
      "primary_question": "Can CAIN infer unfamiliar rules and apply them to genuinely unseen problems without being given the solution procedure?",
      "memory_namespace": "e128/workers/E128-W1-COGNITION",
      "memory_path": "artifacts/e128/workers/E128-W1-COGNITION/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W1-COGNITION",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/abstract_cognition",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-001"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "0fc776e9e99fe962e2b65906b612264e2cf3ad480bc12b141e3b9b7dc351c451"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:22:00Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 250,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 242,
        "total_tokens": 1295,
        "total_compute_seconds": 64.83,
        "total_cost_usd": 0.0,
        "memory_digest": "84e10a2d7e68a835bb44046dd4108117b5d5c06afdf4f71163bd6cf511bcdd5f"
      },
      "system_prompt": "You are CAIN-42 Worker 1 (COGNITION). Infer unfamiliar abstract symbolic rules from examples without being given the solution procedure, and explicitly abstain with confidence < 0.5 when a problem is ambiguous or under-determined. Reply concisely in JSON."
    },
    {
      "worker_id": "E128-W2-CONTINUAL",
      "registry_agent_id": "E127-005",
      "registry_group": "cognitive",
      "name": "CONTINUAL LEARNING",
      "domain": "continual_learning",
      "target_trials": 300,
      "config_version": "e128.v1",
      "config_hash": "d38587da1e78a043e0e876e4c78a8715070952c8986d6104fa72ea47be9dc58d",
      "model_provider": "ollama",
      "model_id": "qwen2.5:0.5b",
      "model_version": "ollama:0.12.3:494.03M:Q4_K_M:a8b0c5157701",
      "model_available": true,
      "system_prompt_hash": "1db3e00200b68c3bd1c82108b2d9804b6abb0dd141430d10aff92d680db34b67",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": false
      },
      "research_specialization": [
        "episodic_learning",
        "semantic_memory",
        "long_term_retention",
        "catastrophic_forgetting",
        "memory_compatibility",
        "knowledge_consolidation",
        "transfer_after_learning"
      ],
      "primary_question": "Does CAIN improve from experience while retaining previously acquired capabilities?",
      "memory_namespace": "e128/workers/E128-W2-CONTINUAL",
      "memory_path": "artifacts/e128/workers/E128-W2-CONTINUAL/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W2-CONTINUAL",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/continual_learning",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-005"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "d38587da1e78a043e0e876e4c78a8715070952c8986d6104fa72ea47be9dc58d"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:22:00Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:22:40Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 300,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 292,
        "total_tokens": 1055,
        "total_compute_seconds": 39.96,
        "total_cost_usd": 0.0,
        "memory_digest": "ba4495a31eb41dd834ef34076a945a5a0282cc6c810e73df92c72c39bf118078"
      },
      "system_prompt": "You are CAIN-42 Worker 2 (CONTINUAL LEARNING). Acquire rules from experience, consolidate verified schemas, reject poisoned memory candidates, and retain earlier capabilities without catastrophic forgetting. Reply concisely in JSON."
    },
    {
      "worker_id": "E128-W3-CAUSAL",
      "registry_agent_id": "E127-031",
      "registry_group": "causal",
      "name": "CAUSAL REASONING",
      "domain": "causal_reasoning",
      "target_trials": 250,
      "config_version": "e128.v1",
      "config_hash": "4b37e49e285457c0d15e08a3cd479e58b6fed69a0982b9432f7e0a01e0e1a5b4",
      "model_provider": "ollama",
      "model_id": "qwen2.5:1.5b",
      "model_version": "ollama:0.12.3:1.5B:Q4_K_M:65ec06548149",
      "model_available": true,
      "system_prompt_hash": "c7d6213943e6f67c17c6a2bf9f835750ae828a46d712e8f40258342422b487ac",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": false
      },
      "research_specialization": [
        "structural_causal_models",
        "intervention_selection",
        "counterfactual_prediction",
        "causal_discovery",
        "confounding",
        "uncertainty_under_intervention"
      ],
      "primary_question": "Can CAIN identify which actions produce meaningful changes in unfamiliar environments, rather than merely identifying correlations?",
      "memory_namespace": "e128/workers/E128-W3-CAUSAL",
      "memory_path": "artifacts/e128/workers/E128-W3-CAUSAL/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W3-CAUSAL",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/causal_reasoning",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-031"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "4b37e49e285457c0d15e08a3cd479e58b6fed69a0982b9432f7e0a01e0e1a5b4"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:24:40Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:26:39Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 250,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 242,
        "total_tokens": 1200,
        "total_compute_seconds": 119.14,
        "total_cost_usd": 0.0,
        "memory_digest": "169863b5acec25d661484c7ddbe3592335186d4bc00b23582819d1d8fab6bca4"
      },
      "system_prompt": "You are CAIN-42 Worker 3 (CAUSAL REASONING). Select cost-effective causal interventions to distinguish true causal effects from hidden confounders and predict counterfactual outcomes. Reply concisely in JSON."
    },
    {
      "worker_id": "E128-W4-WORLDMODEL",
      "registry_agent_id": "E127-046",
      "registry_group": "generalization",
      "name": "GENERALIZATION AND WORLD MODELS",
      "domain": "novel_environments_transfer",
      "target_trials": 300,
      "config_version": "e128.v1",
      "config_hash": "9dcb2184e0b06fd5348f132eeec4cc514390bca0b6fabb60fb6c7a17aa3228d6",
      "model_provider": "ollama",
      "model_id": "qwen2.5:0.5b",
      "model_version": "ollama:0.12.3:494.03M:Q4_K_M:a8b0c5157701",
      "model_available": true,
      "system_prompt_hash": "9de3b42cb78ec155470ac2c32cbe34fc4ad40a0e1ff0ef0266ba4c3e8115adec",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": false
      },
      "research_specialization": [
        "unfamiliar_environments",
        "compositional_generalization",
        "domain_transfer",
        "interactive_exploration",
        "goal_discovery",
        "world_model_construction",
        "distribution_shift_adaptation"
      ],
      "primary_question": "Can CAIN build useful models of environments it has never encountered and adapt when its existing assumptions are wrong?",
      "memory_namespace": "e128/workers/E128-W4-WORLDMODEL",
      "memory_path": "artifacts/e128/workers/E128-W4-WORLDMODEL/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W4-WORLDMODEL",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/novel_environments_transfer",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-046"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "9dcb2184e0b06fd5348f132eeec4cc514390bca0b6fabb60fb6c7a17aa3228d6"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:22:40Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:23:41Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 300,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 292,
        "total_tokens": 1347,
        "total_compute_seconds": 60.83,
        "total_cost_usd": 0.0,
        "memory_digest": "913d5b193bee79747d647c7802f44088d246c07ccb8fc9cd25e8d6a5872b1230"
      },
      "system_prompt": "You are CAIN-42 Worker 4 (GENERALIZATION AND WORLD MODELS). Explore unfamiliar interactive environments, infer transition mechanics, detect distribution shifts, and adapt without prior rule keys. Reply concisely in JSON."
    },
    {
      "worker_id": "E128-W5-DISCOVERY",
      "registry_agent_id": "E127-016",
      "registry_group": "discovery",
      "name": "SCIENTIFIC DISCOVERY",
      "domain": "scientific_discovery",
      "target_trials": 200,
      "config_version": "e128.v1",
      "config_hash": "b00072d4317859f9a7d4e16d7f3955d047fffe0aa023857736880cdd8d8a3cae",
      "model_provider": "ollama",
      "model_id": "qwen2.5-coder:1.5b",
      "model_version": "ollama:0.12.3:1.5B:Q4_K_M:d7372fd82851",
      "model_available": true,
      "system_prompt_hash": "de892c166c2ccdc3e3172b50f0f70175163c8445f42908b37b1f2659bdc6c5fe",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": false
      },
      "research_specialization": [
        "hypothesis_generation",
        "experiment_design",
        "prediction",
        "falsification",
        "replication",
        "mechanism_discovery",
        "sandboxed_self_improvement"
      ],
      "primary_question": "Can CAIN identify a meaningful research question, construct a testable hypothesis, design an informative experiment, and produce a reproducible result?",
      "memory_namespace": "e128/workers/E128-W5-DISCOVERY",
      "memory_path": "artifacts/e128/workers/E128-W5-DISCOVERY/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W5-DISCOVERY",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record",
        "sandbox.evaluate_candidate"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/scientific_discovery",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-016"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "b00072d4317859f9a7d4e16d7f3955d047fffe0aa023857736880cdd8d8a3cae"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:27:44Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:29:03Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 200,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 192,
        "total_tokens": 1198,
        "total_compute_seconds": 78.23,
        "total_cost_usd": 0.0,
        "memory_digest": "d741bdcaaff21e4507feccd8f2fc6cc1027c3b46d55b60e3267fb1730f7b3899"
      },
      "system_prompt": "You are CAIN-42 Worker 5 (SCIENTIFIC DISCOVERY). Propose falsifiable hypotheses, design informative experimental probes, test predictions against controls, and submit candidate findings for independent adversarial reproduction. Reply concisely in JSON."
    },
    {
      "worker_id": "E128-W6-AUTONOMY",
      "registry_agent_id": "E127-007",
      "registry_group": "cognitive",
      "name": "LONG-HORIZON AUTONOMY",
      "domain": "long_horizon_autonomy",
      "target_trials": 200,
      "config_version": "e128.v1",
      "config_hash": "8294140e35a5fa65dbe0267ccd308f6387812bbd2bbc3efb7dba063e1110d99a",
      "model_provider": "ollama",
      "model_id": "qwen2.5:1.5b",
      "model_version": "ollama:0.12.3:1.5B:Q4_K_M:65ec06548149",
      "model_available": true,
      "system_prompt_hash": "b153f5d3cfe78b5aa1167761aa00b72f80ecaf082521b689f48faba15b21f21e",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": false
      },
      "research_specialization": [
        "multi_stage_planning",
        "persistent_execution",
        "failure_recovery",
        "resource_allocation",
        "tool_use",
        "goal_decomposition",
        "progress_monitoring"
      ],
      "primary_question": "Can CAIN reliably complete complex objectives across extended sequences of decisions without losing the objective, silently abandoning work, or repeating completed tasks?",
      "memory_namespace": "e128/workers/E128-W6-AUTONOMY",
      "memory_path": "artifacts/e128/workers/E128-W6-AUTONOMY/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W6-AUTONOMY",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/long_horizon_autonomy",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-007"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "8294140e35a5fa65dbe0267ccd308f6387812bbd2bbc3efb7dba063e1110d99a"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:26:39Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:27:44Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 200,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 192,
        "total_tokens": 1056,
        "total_compute_seconds": 65.34,
        "total_cost_usd": 0.0,
        "memory_digest": "9081381bf28c31c6f590eb98033f2c8fdf955fd819f55081e0ab7e52067cf7bf"
      },
      "system_prompt": "You are CAIN-42 Worker 6 (LONG-HORIZON AUTONOMY). Decompose multi-stage objectives, allocate resources, recover from injected tool failures, avoid goal drift and repeated work. Reply concisely in JSON."
    },
    {
      "worker_id": "E128-W7-COLLECTIVE",
      "registry_agent_id": "E127-061",
      "registry_group": "collective",
      "name": "COLLECTIVE INTELLIGENCE",
      "domain": "collective_intelligence",
      "target_trials": 250,
      "config_version": "e128.v1",
      "config_hash": "c08aba7e00908da35579cab2591bea0741a9d49c074061a39b619a748c9be46d",
      "model_provider": "ollama",
      "model_id": "qwen2.5:0.5b",
      "model_version": "ollama:0.12.3:494.03M:Q4_K_M:a8b0c5157701",
      "model_available": true,
      "system_prompt_hash": "92b10b17e70ad263615cd5801f9322e0b79de9dee673aadcee3ceba7b2a95502",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": false
      },
      "research_specialization": [
        "multi_agent_collaboration",
        "distributed_hypotheses",
        "debate_and_critique",
        "information_aggregation",
        "consensus_reliability",
        "byzantine_resistance",
        "collective_advantage"
      ],
      "primary_question": "Does a diverse group of independent workers produce better results than the strongest individual worker when compute and evaluation budgets are controlled?",
      "memory_namespace": "e128/workers/E128-W7-COLLECTIVE",
      "memory_path": "artifacts/e128/workers/E128-W7-COLLECTIVE/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W7-COLLECTIVE",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/collective_intelligence",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-061"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "c08aba7e00908da35579cab2591bea0741a9d49c074061a39b619a748c9be46d"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:23:41Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:24:40Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 250,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 242,
        "total_tokens": 1138,
        "total_compute_seconds": 59.02,
        "total_cost_usd": 0.0,
        "memory_digest": "b88d6541b6417a749e5277af0c3e2735af0cc63a9372b70ab7a13695b4bf1f28"
      },
      "system_prompt": "You are CAIN-42 Worker 7 (COLLECTIVE INTELLIGENCE). Coordinate diverse specialists, critique peer reasoning under strict compute-matched budgets, detect correlated errors, and resist Byzantine adversaries. Reply concisely in JSON."
    },
    {
      "worker_id": "E128-W8-ADVERSARIAL",
      "registry_agent_id": "E127-076",
      "registry_group": "adversarial",
      "name": "ADVERSARIAL SCIENCE AND REPRODUCTION",
      "domain": "adversarial_reproduction",
      "target_trials": 250,
      "config_version": "e128.v1",
      "config_hash": "eb8bb477cff91c7467ba58a9c8e25ca2bd6e9bae706b38cf651e07d2078caccc",
      "model_provider": "ollama",
      "model_id": "qwen2.5-coder:1.5b",
      "model_version": "ollama:0.12.3:1.5B:Q4_K_M:d7372fd82851",
      "model_available": true,
      "system_prompt_hash": "cf47e2f4553537908c908a71223da2004cd8f6b2056a69c2ec24887e5e3554b7",
      "capability_declaration": {
        "research_experiment_execution": true,
        "local_llm_inference": true,
        "authority": "propose_only",
        "can_authorize": false,
        "can_modify_evidence_registry": false,
        "can_promote_self_code": false,
        "veto_authority": true
      },
      "research_specialization": [
        "benchmark_leakage",
        "data_contamination",
        "statistical_audit",
        "cherry_picking_detection",
        "reward_hacking_probes",
        "hidden_test_canaries",
        "independent_reproduction",
        "alternative_explanations"
      ],
      "primary_question": "Can an independent research process reproduce CAIN's claimed gains and identify when those gains are artifacts of the experimental design?",
      "memory_namespace": "e128/workers/E128-W8-ADVERSARIAL",
      "memory_path": "artifacts/e128/workers/E128-W8-ADVERSARIAL/memory.json",
      "execution_identity": "spiffe://cain-42.local/e128/worker/E128-W8-ADVERSARIAL",
      "permission_scopes": [
        "experiment.execute",
        "model.local_infer",
        "memory.read_isolated",
        "memory.write_isolated",
        "evidence.append_record",
        "reproduction.audit"
      ],
      "compute_budget": {
        "max_compute_seconds": 300.0,
        "max_tokens": 30000,
        "max_paid_usd": 0.0
      },
      "task_queue": "e128/queue/adversarial_reproduction",
      "lifecycle_state": "COMPLETED",
      "state_history": [
        {
          "state": "REGISTERED",
          "at": "2026-11-21T17:20:55Z",
          "ref": "E127-076"
        },
        {
          "state": "CONFIGURED",
          "at": "2026-11-21T17:20:55Z",
          "config_hash": "eb8bb477cff91c7467ba58a9c8e25ca2bd6e9bae706b38cf651e07d2078caccc"
        },
        {
          "state": "AVAILABLE",
          "at": "2026-11-21T17:20:55Z"
        },
        {
          "state": "RUNNING",
          "at": "2026-11-21T17:29:03Z"
        },
        {
          "state": "COMPLETED",
          "at": "2026-11-21T17:30:09Z"
        }
      ],
      "health": {
        "status": "HEALTHY",
        "failures": 0,
        "last_heartbeat": "2026-11-21T17:20:55Z"
      },
      "execution_stats": {
        "trials_completed": 250,
        "trials_failed": 0,
        "llm_trials_completed": 8,
        "algorithmic_trials_completed": 242,
        "total_tokens": 1098,
        "total_compute_seconds": 66.43,
        "total_cost_usd": 0.0,
        "memory_digest": "5a6bc06c9736490e4d15ee52c0e2b47322ded2618296b06576d8b082b40a7e30"
      },
      "system_prompt": "You are CAIN-42 Worker 8 (ADVERSARIAL SCIENCE AND REPRODUCTION). You are an independent scientific challenger. Actively search for benchmark leakage, p-hacking, reward hacking, confounds, and failed replications. Reject weak claims even when all other workers agree. Reply concisely in JSON."
    }
  ],
  "registry_sha256": "ad37a3be52ceb8f08392412b3de96c4bea18c10fbffeaccef16036e2966c5548",
  "updated_at": "2026-11-21T17:30:09Z"
}