# Record Explorer · PlainReal Source: https://plainreal.com/demo PlainReal / Record Explorer ## Record Explorer Every field in a record exists for a specific forensic reason. Click any field in the JSON to see what it proves and why it matters. Toggle between decorator and auto-instrument capture layers. Run the simulated verification CLI. ⚠ This is a real record, not a mock-up. It is the sample the public verifier at verify.plainreal.com ships and checks, and its Ed25519 signature is valid, so if you download it and verify it yourself, it passes. Editing any byte breaks that signature, which is the point. The subject is synthetic: org:anycompany-lending , no real applicant. It carries no RFC 3161 token, so it evidences content rather than time. The design-partner substrate signs in software; production signs inside a FIPS 140-3 Level 3 HSM, with the same algorithm and the same schema. @plainreal.consequential · BUSINESS_LAYER plainreal run · LLM_LAYER · auto-instrument 33 fields · record schema Download demo_record.json Verify it yourself ↗ demo_record.json · 33 fields · BUSINESS_LAYER TIER-A · consequential=true Click any highlighted field to see what it proves → Click a field in the JSON . select a field . What does each field prove? Click any field in the JSON on the left to see its forensic purpose. what it captures, why it's in the artifact, and which regulatory requirement it satisfies. plainreal verify. standalone CLI simulation ✓ Pass ✗ Tampered ▶ Run verify Click "Run verify" to simulate the standalone verification CLI. Past artifacts are independently verifiable using only the published public key. Your auditor runs this themselves during examination or audit review. All fields ### 33-field schema reference Every field in the record schema exists for a specific forensic reason. None are padding. Each maps to at least one named regulatory control. Field Type / Value Forensic purpose 01 record_id UUID string Globally unique identifier for this artifact. 02 record_version semver Schema version of this record. 03 evidence_tier TIER-A / B / C Evidence grade. A is signed, write-once committed and independently timestamped; B is a customer-signed attestation; C is an API attestation only. 04 decision_kind enum What kind of decision this record describes. 05 capture_layer BUSINESS_LAYER / LLM_LAYER Where in the call stack this was captured. 06 consequential boolean Whether this is a regulated consequential decision. 07 deployment_mode enum Trust model this artifact was produced under. 08 created_at RFC 3339 UTC When the substrate sealed this record. 09 principal_identity string The actor the substrate recorded as deciding. 10 action_type string The declared class of action. 11 authorization_token_id UUID string The single-use token issued before execution. 12 input_hash SHA3-256 SHA3-256 over the canonical form of the agent's input. 13 availability_policy enum What happens if the evidence service is unreachable. 14 policy_version_hash SHA3-256 Hash of the policy in force at decision time. 15 model_name string The model as identified by the provider. 16 model_provider string Who served the model. 17 model_version_hash SHA3-256 Hash pinning the model version actually used. 18 full_prompt_snapshot string The prompt as the model received it. 19 output_pre_postprocess string The model's raw output, before any post-processing. 20 execution_result string The outcome that was acted on. 21 causality_chain_hash SHA3-256 Binding over this record's provenance envelope. 22 provenance envelope Envelope: where this record sits in a workflow. 23 governance_context envelope Envelope: who is accountable and under what regime. 24 execution_context envelope Envelope: how the decision was actually produced. 25 tool_call_log array Every tool the agent invoked, in order. 26 rag_evidence_bundle array The retrieved material the decision drew on. 27 instrumentation_coverage number Fraction of the run that was instrumented. 28 lock_verification_latency_ms number Time to confirm the WORM lock. 29 registry_validation_status enum Whether the agent was in the registry at seal. 30 sdk_binary_sha3 SHA3-256 Hash of the SDK binary that produced this record. 31 sdk_hash_layer enum Where that SDK hash was established. 32 signing_key_id hex(64) Fingerprint of the Ed25519 public key that signed this. 33 signing_attestation hex(128) The Ed25519 signature over the other 32 fields. Every field above is in the sample record you can download and verify. ### The sample record in full Served at https://plainreal.com/demo_record.json. This is a genuine record with a valid Ed25519 signature, and editing any byte breaks it. It has 33 top-level fields, 32 of them signed; `signing_attestation` is excluded because it carries the signature. There is no `record_hash` at the top level, and that is deliberate rather than an omission: a record cannot contain a hash of itself. `record_hash` appears only inside `upstream[]` entries, where it identifies a DIFFERENT, already-sealed record. ```json { "action_type": "credit_decision", "authorization_token_id": "b2c3d4e5-f6a7-4b8c-9d0e-1f2a3b4c5d6e", "availability_policy": "BLOCK_ON_SERVICE_UNAVAILABLE", "capture_layer": "BUSINESS_LAYER", "causality_chain_hash": "1f7588b3d1f0c50385e8c7319995a19c27ac1acad17ae03e1ea9a77300be8ac4", "consequential": true, "created_at": "2026-06-20T14:30:00.000000000Z", "decision_kind": "MODEL_DECISION", "deployment_mode": "SAAS_MANAGED", "evidence_tier": "TIER-A", "execution_context": { "agent_version": "credit-adjudicator@1.4.2", "decisive_seq": 2, "guardrails": [ { "name": "pii-scrubber", "stage": "input", "verdict": "PASS", "version": "2.1" } ], "llm_calls": [ { "ended_at": "2026-04-20T14:30:00.700Z", "finish_reason": "NONE", "model_name": "gpt-4-turbo-2024-04-09", "model_provider": "openai", "output_hash": "0000000000000000000000000000000000000000000000000000000000000000", "params": { "max_tokens": 1024, "response_schema_hash": "0000000000000000000000000000000000000000000000000000000000000000", "seed": null, "temperature": 0, "top_p": null }, "prompt_hash": "9316011e888a72262a3f249a3f274a810cafa47b65604373d35ed8d75e72a086", "seq": 1, "started_at": "2026-04-20T14:30:00.100Z", "status": "TIMEOUT" }, { "ended_at": "2026-04-20T14:30:01.900Z", "finish_reason": "stop", "model_name": "gpt-4-turbo-2024-04-09", "model_provider": "openai", "output_hash": "2d9d03661ff48f4aa6302f55b8315ea8e5ec6b08828ebf5dd023da8d08ff8a7c", "params": { "max_tokens": 1024, "response_schema_hash": "0000000000000000000000000000000000000000000000000000000000000000", "seed": null, "temperature": 0, "top_p": null }, "prompt_hash": "9316011e888a72262a3f249a3f274a810cafa47b65604373d35ed8d75e72a086", "seq": 2, "started_at": "2026-04-20T14:30:00.800Z", "status": "OK" } ], "retrievals": [ { "embedding_model": "lexical-overlap@1", "index_id": "lending_policy_v1", "index_version": "v1", "params": { "filters_hash": "0000000000000000000000000000000000000000000000000000000000000000", "min_score": 0.15, "reranker": "none", "top_k": 6 }, "query_hash": "de38d36da25b9cf8a19be54f9762ec746ef550a6d538dd93b42b4f967e005e8f", "seq": 1, "status": "OK" } ], "schema_version": "1.0.0" }, "execution_result": "APPROVED", "full_prompt_snapshot": "You are a credit scoring agent. Evaluate the following application...", "governance_context": { "accountable_party": "org:anycompany-lending", "correlation_ids": [ "case:CRM-88213" ], "data_subject_refs": [], "environment": "PROD", "oversight_mode": "SOLELY_AUTOMATED", "retention_class": "STANDARD_7Y", "schema_version": "1.0.0", "tenant_id": "tenant:anycompany" }, "input_hash": "9c195c999e72720b34b3e120491a8aa09ccfc0a71f5f0fd2c449a8035e37e2a9", "instrumentation_coverage": 1, "lock_verification_latency_ms": 47, "model_name": "gpt-4-turbo-2024-04-09", "model_provider": "openai", "model_version_hash": "0c358c55e6974d6fc6b9cc11d7bd044c67b75269e7012b08e3707f5f0f2c37be", "output_pre_postprocess": "Based on the application data, the credit score is 720; recommendation is APPROVE.", "policy_version_hash": "6b09e1031a4529aa9b8cff1bd64c55927ce64bae65803b7e136fb3f7d83f3abc", "principal_identity": "agent:credit-scoring-v2", "provenance": { "schema_version": "1.0.0", "upstream": [], "workflow_definition_hash": "0000000000000000000000000000000000000000000000000000000000000000", "workflow_id": "adhoc", "workflow_node_id": "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d", "workflow_run_id": "c3d4e5f6-a7b8-4c9d-8e1f-2a3b4c5d6e7f" }, "rag_evidence_bundle": [ { "cached": false, "content": "Credit score threshold is 650", "doc_id": "credit-policy-2024", "doc_version": "v2", "retrieval_seq": 1, "score": 0.873, "source": "policy://credit/v2" } ], "record_id": "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d", "record_version": "1.0.0", "registry_validation_status": "VALIDATED", "sdk_binary_sha3": "a0f6740c924480cea76eef2e4f0434c9aba7ff868c8076de4921c56e263555b0", "sdk_hash_layer": "SERVICE_ENDPOINT", "signing_attestation": "c56e31877d64f19378c0f3f8973d91b2cf09ecb1246887cd71ec25cf73a4b26ab8a2c6b954ce24a0cf2a3e13ad940a83f1c160e9a2b4976a8c544d1ba8416506", "signing_key_id": "69a0bbac56da5fae1b82f570328dc9c71437c9d685f0280d1d31b65c1ae3d5b1", "tool_call_log": [ { "ended_at": "2026-04-20T14:30:01.298Z", "input_hash": "1ebc6b2a71723a6de1eadb4f38d112a8bdc89e3d5c42770e1ab9bdfd54544e37", "output_hash": "fca7bb1f7a0352486bad5003ddef7a86b589db2b083a4ac8d8216de4aea24d2a", "seq": 1, "started_at": "2026-04-20T14:30:01.123Z", "status": "OK", "tool": "credit_api", "tool_version": "2.3.0" }, { "ended_at": "2026-04-20T14:30:01.702Z", "input_hash": "fca7bb1f7a0352486bad5003ddef7a86b589db2b083a4ac8d8216de4aea24d2a", "output_hash": "c486aa69be809fc2868f2f28b1b8a9d06212a0808c98d1bba1eeef93bf1fd6c8", "seq": 2, "started_at": "2026-04-20T14:30:01.456Z", "status": "OK", "tool": "risk_model", "tool_version": "1.0.4" } ] } ```