This document summarizes the MCP tool surface and expected usage. Full schemas are in tooling.json, with supporting schemas under schemas/ and examples under examples/.
Lifecycle quickstart
scenario_definevalidates and registers a monotone-DAG scenario law.scenario_startcreates an empty-progress run bound to that exact law.scenario_statusandscenario_historyread accepted state and records.scenario_open_stageexplicitly opens one ready stage.scenario_evaluate_stagerecords an attempt and may complete that stage.scenario_precheck_stagepredicts evaluation without mutation.scenario_submitcommits a run-wide or exact stage-scoped artifact.runpack_exportandrunpack_verifysupport offline verification.
Artifact references
authoring.md: authoring formats and normalization guidance.Docs/generated/decision-gate/examples/scenario.json: hostile RawScenarioSpec example.Docs/generated/decision-gate/examples/scenario.ron: authoring-friendly RawScenarioSpec example.Docs/generated/decision-gate/examples/decision-gate.toml: MCP and bounded local-acquisition config example.
| Tool | Description |
|---|---|
| scenario_define | Admit hostile monotone-DAG scenario syntax and return exact scenario-law and graph identities. |
| scenario_start | Create one empty graph-bound run under the accepted-run authority and return its minted RunId. |
| scenario_status | Return the exact accepted head, automatic run status, stage classifications, and ready frontier. |
| scenario_history | Return bounded accepted records in exact run-head order. |
| scenario_open_stage | Explicitly open one ReadyUnopened stage at an exact accepted head. |
| scenario_evaluate_stage | Acquire evidence and evaluate one OpenIncomplete stage at an exact accepted head. |
| scenario_precheck_stage | Predict one stage RET result from asserted evidence without opening or mutation. |
| scenario_submit | Accept a run-wide or exact stage-open-scoped artifact at one exact run head. |
| scenario_resolve_creation | Resolve a potentially indeterminate creation through the accepted creation-receipt authority. |
| scenario_resolve_mutation | Resolve a potentially indeterminate mutation through its exact accepted receipt. |
| runpack_export | Publish a coherent accepted-prefix runpack-v6 subject locally. |
| runpack_verify | Verify one supplied runpack-v6 subject with a total claim report. |
| schemas_register | Register a data shape schema for a namespace. |
| schemas_list | List registered data shapes for a namespace. |
| schemas_get | Fetch a specific data shape by identifier and version. |
| scenarios_list | List registered scenarios for a namespace. |
| decision_gate_docs_search | Search Decision Gate documentation for runtime guidance. |
scenario_define
Admit hostile monotone-DAG scenario syntax and return exact scenario-law and graph identities.
Inputs
spec(required): Hostile scenario specification to validate and register.
Outputs
graph_identity(required): Canonical DG projection of a sealed Xarxa graph identity.scenario_id(required): Scenario identifier.scenario_law_identity(required): Domain-separated canonical Decision Gate scenario-law identity.
Notes
- Use before starting runs; scenario_id and scenario_law_identity bind later calls.
- Validates Xarxa topology, RET proofs, typed conditions, acquisition compatibility, and effective value limits.
- Canonical law and graph identities are deterministic under raw ordering changes.
- Fails closed on invalid specs or duplicate registration of the same exact law; distinct immutable revisions remain addressable.
Example
Register the example scenario spec.
Input:
{
"spec": {
"acquisition_bindings": [],
"completion_requirement": {
"Condition": "approve"
},
"conditions": [
{
"condition_id": "environment_is_production",
"evidence_use_policy": {
"allowed_channels": [
"caller_submission"
],
"authorization": "embedding_authority",
"combination": {
"combination": "AllQualifyingAgree"
},
"freshness": {
"freshness": "NotRequired"
},
"independence": "distinct_acquisition",
"integrity": "canonical_content_only",
"replay": "full_semantic",
"required_assurance": [
"content_integrity",
"scope",
"schema_compatibility"
],
"retention": "retain_canonical_value",
"scope": "exact_condition"
},
"observation_domain": {
"domain": "String",
"maximum_bytes": 64,
"minimum_bytes": 1
},
"predicate": {
"expected": {
"value": "production",
"value_type": "String"
},
"predicate": "Equals"
}
},
{
"condition_id": "after_freeze",
"evidence_use_policy": {
"allowed_channels": [
"caller_submission"
],
"authorization": "embedding_authority",
"combination": {
"combination": "AllQualifyingAgree"
},
"freshness": {
"freshness": "NotRequired"
},
"independence": "distinct_acquisition",
"integrity": "canonical_content_only",
"replay": "full_semantic",
"required_assurance": [
"content_integrity",
"scope",
"schema_compatibility"
],
"retention": "retain_canonical_value",
"scope": "exact_condition"
},
"observation_domain": {
"domain": "Boolean"
},
"predicate": {
"expected": {
"value": true,
"value_type": "Boolean"
},
"predicate": "Equals"
}
}
],
"namespace_id": 1,
"policies": [],
"scenario_id": "example-scenario",
"schemas": [],
"spec_version": "dag-v3",
"stages": [
{
"completion_requirement": {
"Condition": "environment_is_production"
},
"entry_packets": [],
"prerequisite": {
"kind": "root"
},
"stage_id": "inspect"
},
{
"completion_requirement": {
"Condition": "after_freeze"
},
"entry_packets": [],
"prerequisite": {
"kind": "requires",
"requirement": {
"Condition": "inspect"
}
},
"stage_id": "approve"
}
],
"value_limits": {
"selection": "InheritOperator"
}
}
}
Output:
{
"graph_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"scenario_id": "example-scenario",
"scenario_law_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
}
scenario_start
Create one empty graph-bound run under the accepted-run authority and return its minted RunId.
Inputs
correlation(required): One of: object, object.creation_request_id(required): Stable run-creation request identity.genesis(required): Type: object.namespace_id(required): Type: integer.scenario_id(required): Scenario identifier.scenario_law_identity(required): Domain-separated canonical Decision Gate scenario-law identity.
Outputs
outcome(required): One of: object, object, object, object, object, object, object.
Notes
- Requires namespace_id, exact scenario_law_identity, and a stable 128-bit creation_request_id.
- RunId and accepted audit time are minted by named authorities; callers supply neither.
- Start never opens roots or proposes entry intents.
- Accepted creation replay is receipt-first and returns the exact original result.
Example
Create empty progress and receive an authority-minted RunId.
Input:
{
"correlation": {
"kind": "unset"
},
"creation_request_id": "10000000000000000000000000000001",
"genesis": {
"kind": "empty"
},
"namespace_id": 1,
"scenario_id": "example-scenario",
"scenario_law_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
}
Output:
{
"outcome": {
"outcome": "accepted",
"result": {
"acceptance_time": {
"accepted_audit_instant": 0,
"authority_revision": "dg-local-system-wall-clock-revision-v1",
"source_material_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"time_authority_id": "dg-local-system-wall-clock-v1"
},
"admission": {
"kind": "embedded",
"profile_id": "dg-embedded-authority-v1",
"request_provenance_digest": {
"algorithm": "sha256",
"value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
}
},
"correlation": {
"kind": "unset"
},
"creation_request_id": "10000000000000000000000000000001",
"genesis": {
"kind": "empty"
},
"head": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
},
"invocation_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"run_key": {
"namespace_id": 1,
"run_id": "00000000000000000000000000000001"
},
"scenario_law_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
}
}
}
scenario_status
Return the exact accepted head, automatic run status, stage classifications, and ready frontier.
Inputs
namespace_id(required): Type: integer.run_id(required): Authority-minted run identity.scenario_id(required): Scenario identifier.
Outputs
accepted_head(required): Type: object.committed_intent_count(required): Total committed intent count without unbounded identity materialization.ready_frontier(required): Type: array.recent_attempts(required): Type: array.stages(required): Type: array.status(required): Type: string.
Notes
- Read-only status does not open, evaluate, or mutate a stage.
- ReadyUnopened and OpenIncomplete are distinct and carry no assignment semantics.
Example
Poll run status without advancing the run.
Input:
{
"namespace_id": 1,
"run_id": "00000000000000000000000000000001",
"scenario_id": "example-scenario"
}
Output:
{
"accepted_head": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
},
"committed_intent_count": 0,
"ready_frontier": [
"inspect"
],
"recent_attempts": [],
"stages": [
{
"classification": "ready_unopened",
"stage_id": "inspect"
}
],
"status": "active"
}
scenario_history
Return bounded accepted records in exact run-head order.
Inputs
after_revision(optional, nullable): One of: null, integer.limit(optional): Type: integer.namespace_id(required): Type: integer.run_id(required): Authority-minted run identity.scenario_id(required): Scenario identifier.
Outputs
next_after_revision(required, nullable): One of: null, integer.records(required): Type: array.
Notes
- History is an accepted-order coordination and audit surface, not stage ownership.
Example
Read bounded accepted history.
Input:
{
"after_revision": null,
"limit": 50,
"namespace_id": 1,
"run_id": "00000000000000000000000000000001",
"scenario_id": "example-scenario"
}
Output:
{
"next_after_revision": null,
"records": []
}
scenario_open_stage
Explicitly open one ReadyUnopened stage at an exact accepted head.
Inputs
expected_head(required): Type: object.namespace_id(required): Type: integer.operation_id(required): Stable stage-open operation identity.run_id(required): Authority-minted run identity.scenario_id(required): Scenario identifier.stage_id(required): Explicit stage identifier.
Outputs
outcome(required): One of: object, object, object, object, object, object, object, object.
Notes
- Opening records run-level work start and proposes entry intents once; it does not assign or exclude siblings.
Example
Open one ready stage at an exact head.
Input:
{
"expected_head": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
},
"namespace_id": 1,
"operation_id": "20000000000000000000000000000001",
"run_id": "00000000000000000000000000000001",
"scenario_id": "example-scenario",
"stage_id": "inspect"
}
Output:
{
"outcome": {
"outcome": "stale",
"result": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
}
}
}
scenario_evaluate_stage
Acquire evidence and evaluate one OpenIncomplete stage at an exact accepted head.
Inputs
caller_evidence(optional): Type: array.expected_head(required): Type: object.local_acquisition(optional): Type: array.namespace_id(required): Type: integer.operation_id(required): Stable stage-evaluation operation identity.run_id(required): Authority-minted run identity.scenario_id(required): Scenario identifier.stage_id(required): Explicit stage identifier.
Outputs
outcome(required): One of: object, object, object, object, object, object, object, object.
Notes
- False and Unknown append retryable attempts; only True proposes stage completion.
- Operational acquisition or evaluator failure mints no semantic proposal.
Example
Evaluate one open stage at an exact head.
Input:
{
"caller_evidence": [
{
"condition_id": "environment_is_production",
"retention": {
"retention": "RetainCanonicalValue"
},
"value": {
"value": "production",
"value_type": "String"
}
}
],
"expected_head": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
},
"local_acquisition": [],
"namespace_id": 1,
"operation_id": "30000000000000000000000000000001",
"run_id": "00000000000000000000000000000001",
"scenario_id": "example-scenario",
"stage_id": "inspect"
}
Output:
{
"outcome": {
"outcome": "stale",
"result": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
}
}
}
scenario_precheck_stage
Predict one stage RET result from asserted evidence without opening or mutation.
Inputs
caller_evidence(optional): Type: array.local_acquisition(optional): Type: array.namespace_id(required): Type: integer.scenario_id(required): Scenario identifier.scenario_law_identity(required): Domain-separated canonical Decision Gate scenario-law identity.stage_id(required): Stage identifier.
Outputs
outcome(required): Type: string.trace(required): Type: array.
Notes
- Precheck never claims accepted progress or issues entry intents.
Example
Predict one stage result without mutation.
Input:
{
"caller_evidence": [
{
"condition_id": "environment_is_production",
"retention": {
"retention": "RetainCanonicalValue"
},
"value": {
"value": "production",
"value_type": "String"
}
}
],
"local_acquisition": [],
"namespace_id": 1,
"scenario_id": "example-scenario",
"scenario_law_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"stage_id": "inspect"
}
Output:
{
"outcome": "true",
"trace": []
}
scenario_submit
Accept a run-wide or exact stage-open-scoped artifact at one exact run head.
Inputs
content_type(required): Submitted content type.expected_head(required): Type: object.namespace_id(required): Type: integer.operation_id(required): Stable run-submission operation identity.payload(required): One of: object, object, object.run_id(required): Authority-minted run identity.scenario_id(required): Scenario identifier.scope(required): One of: object, object.
Outputs
outcome(required): One of: object, object, object, object, object, object, object, object.
Notes
- Payload is canonically hashed and stored as an immutable accepted submission.
- Payload is persisted in run state/runpack logs; do not send raw secrets.
- Submission never completes or opens a stage by itself.
- Stage scope must bind both stage_id and its exact accepted stage_open_record_id.
- Use for artifacts the model or operator supplies.
Example
Submit a run-wide artifact at an exact accepted head.
Input:
{
"content_type": "application/json",
"expected_head": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
},
"namespace_id": 1,
"operation_id": "40000000000000000000000000000001",
"payload": {
"kind": "json",
"value": {
"artifact": "attestation",
"status": "approved"
}
},
"run_id": "00000000000000000000000000000001",
"scenario_id": "example-scenario",
"scope": {
"kind": "run_wide"
}
}
Output:
{
"outcome": {
"outcome": "stale",
"result": {
"record_id": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"revision": 1
}
}
}
scenario_resolve_creation
Resolve a potentially indeterminate creation through the accepted creation-receipt authority.
Inputs
creation_request_id(required): Run creation request identity.invocation_identity(required): Canonical invocation identity.namespace_id(required): Type: integer.
Outputs
outcome(required): One of: object, object, object, object, object.
Notes
- Authorization is checked before receipt disclosure.
- Resolution never creates an accepted run or resamples acceptance time.
Example
Resolve one creation invocation after an indeterminate response.
Input:
{
"creation_request_id": "10000000000000000000000000000001",
"invocation_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"namespace_id": 1
}
Output:
{
"outcome": {
"outcome": "known_unaccepted",
"result": {
"reason": "no_accepted_receipt"
}
}
}
scenario_resolve_mutation
Resolve a potentially indeterminate mutation through its exact accepted receipt.
Inputs
invocation_identity(required): Canonical invocation identity.namespace_id(required): Type: integer.operation_id(required): One of: object, object, object.run_id(required): Authority-minted run identity.
Outputs
outcome(required): One of: object, object, object, object, object.
Notes
- Authorization is checked before receipt disclosure.
- Resolution is receipt-first and never compares or advances the current head.
Example
Resolve one stage-open invocation after an indeterminate response.
Input:
{
"invocation_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"namespace_id": 1,
"operation_id": {
"family": "open_stage",
"identity": "20000000000000000000000000000001"
},
"run_id": "00000000000000000000000000000001"
}
Output:
{
"outcome": {
"outcome": "known_unaccepted",
"result": {
"reason": "no_accepted_receipt"
}
}
}
runpack_export
Publish a coherent accepted-prefix runpack-v6 subject locally.
Inputs
namespace_id(required): Namespace identifier.run_id(required): Authority-minted run identity.
Outputs
publication(required): One of: object, object, object, object, object, object.
Notes
- The composition root owns the capability-rooted local immutable repository.
- The manifest is published last and contains no clock, path, URI, or mutable metadata.
- Concurrent later acceptance cannot change the exported historical checkpoint.
Example
Publish one coherent accepted prefix to the managed local repository.
Input:
{
"namespace_id": 1,
"run_id": "00000000000000000000000000000001"
}
Output:
{
"publication": {
"outcome": "published",
"runpack_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
}
}
runpack_verify
Verify one supplied runpack-v6 subject with a total claim report.
Inputs
budget(optional, nullable): One of: null, object.profile(optional): Type: string.runpack_identity(required): Exact runpack-v6 subject identity.
Outputs
report(required): Type: object.
Notes
- Uses supplied_offline_material_v1 only; no network, provider, clock, or resolver is consulted.
- Authorization precedes report or claim disclosure.
- External authenticity, currentness, rollback, effect, and deployment claims remain unavailable without separately qualified evidence.
Example
Verify supplied local material with the exact integrity-only claim profile.
Input:
{
"profile": "integrity_only_v1",
"runpack_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
}
Output:
{
"report": {
"assumptions": [
"supplied bytes are the complete material universe for this invocation"
],
"budget": {
"bytes_read": 1073741824,
"canonical_work_bytes": 4294967296,
"hashing_workers": 4,
"objects": 65536,
"records": 1000000,
"tree_nodes": 262144
},
"claims": [
{
"claim": "structural_format_validity",
"status": {
"method": "manifest_constructor_v6",
"status": "satisfied"
}
},
{
"claim": "runpack_subject_identity",
"status": {
"method": "manifest_constructor_v6",
"status": "satisfied"
}
},
{
"claim": "artifact_tree_integrity",
"status": {
"method": "object_tree_traversal_v6",
"status": "satisfied"
}
},
{
"claim": "scenario_law_construction",
"status": {
"status": "not_requested"
}
},
{
"claim": "scenario_graph_binding",
"status": {
"status": "not_requested"
}
},
{
"claim": "run_allocation_integrity",
"status": {
"status": "not_requested"
}
},
{
"claim": "creation_receipt_integrity",
"status": {
"status": "not_requested"
}
},
{
"claim": "accepted_history_continuity",
"status": {
"status": "not_requested"
}
},
{
"claim": "exact_head_reconstruction",
"status": {
"status": "not_requested"
}
},
{
"claim": "current_state_equivalence",
"status": {
"status": "not_requested"
}
},
{
"claim": "operation_receipt_consistency",
"status": {
"status": "not_requested"
}
},
{
"claim": "stage_open_consistency",
"status": {
"status": "not_requested"
}
},
{
"claim": "evaluation_attempt_consistency",
"status": {
"status": "not_requested"
}
},
{
"claim": "submission_consistency",
"status": {
"status": "not_requested"
}
},
{
"claim": "intent_consistency",
"status": {
"status": "not_requested"
}
},
{
"claim": "evidence_structural_admission",
"status": {
"status": "not_requested"
}
},
{
"claim": "evidence_retention_reconstruction",
"status": {
"status": "not_requested"
}
},
{
"claim": "evidence_use_policy_satisfaction",
"status": {
"status": "not_requested"
}
},
{
"claim": "pure_semantic_replay",
"status": {
"status": "not_requested"
}
},
{
"claim": "acquisition_admission_witness_reconstruction",
"status": {
"status": "not_requested"
}
},
{
"claim": "attribution_authorization_availability",
"status": {
"status": "not_requested"
}
},
{
"claim": "acceptance_time_witness_structure",
"status": {
"status": "not_requested"
}
},
{
"claim": "source_authenticity",
"status": {
"status": "not_requested"
}
},
{
"claim": "currentness",
"status": {
"status": "not_requested"
}
},
{
"claim": "rollback_resistance",
"status": {
"status": "not_requested"
}
},
{
"claim": "dispatch_recovery",
"status": {
"status": "not_requested"
}
},
{
"claim": "effect_recovery",
"status": {
"status": "not_requested"
}
},
{
"claim": "deployment_qualification",
"status": {
"status": "not_requested"
}
},
{
"claim": "verifier_build_procedure_provenance",
"status": {
"status": "not_requested"
}
}
],
"material_resolution": "supplied_offline_material_v1",
"profile": "integrity_only_v1",
"report_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"request_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"runpack_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21",
"trusted_bases": [
"sha256",
"serde_jcs_rfc8785"
],
"verifier_procedure": "decision_gate_offline_verifier_v1"
}
}
schemas_register
Register a data shape schema for a namespace.
Inputs
record(required): Type: object.
Outputs
record(required): Type: object.
Notes
- Schemas are immutable; registering the same version twice fails.
- Provide created_at to record when the schema was authored.
Example
Register a data shape schema.
Input:
{
"record": {
"created_at": {
"kind": "unix_millis",
"value": 1710000000000
},
"description": "Asserted payload schema.",
"namespace_id": 1,
"schema": {
"additionalProperties": false,
"properties": {
"deploy_env": {
"type": "string"
}
},
"required": [
"deploy_env"
],
"type": "object"
},
"schema_id": "asserted_payload",
"version": "v1"
}
}
Output:
{
"record": {
"created_at": {
"kind": "unix_millis",
"value": 1710000000000
},
"description": "Asserted payload schema.",
"namespace_id": 1,
"schema": {
"additionalProperties": false,
"properties": {
"deploy_env": {
"type": "string"
}
},
"required": [
"deploy_env"
],
"type": "object"
},
"schema_id": "asserted_payload",
"version": "v1"
}
}
schemas_list
List registered data shapes for a namespace.
Inputs
cursor(optional, nullable): One of: null, string.limit(optional): Maximum number of records to return.namespace_id(required): Namespace identifier.
Outputs
items(required): Type: array.next_token(required, nullable): One of: null, string.
Notes
- Requires namespace_id.
- Supports pagination via cursor + limit.
Example
List data shapes for a namespace.
Input:
{
"cursor": null,
"limit": 50,
"namespace_id": 1
}
Output:
{
"items": [
{
"created_at": {
"kind": "unix_millis",
"value": 1710000000000
},
"description": "Asserted payload schema.",
"namespace_id": 1,
"schema": {
"additionalProperties": false,
"properties": {
"deploy_env": {
"type": "string"
}
},
"required": [
"deploy_env"
],
"type": "object"
},
"schema_id": "asserted_payload",
"version": "v1"
}
],
"next_token": null
}
schemas_get
Fetch a specific data shape by identifier and version.
Inputs
namespace_id(required): Namespace identifier.schema_id(required): Data shape identifier.version(required): Data shape version identifier.
Outputs
record(required): Type: object.
Notes
- Requires namespace_id, schema_id, and version.
- Fails closed when schema is missing.
Example
Fetch a data shape by identifier and version.
Input:
{
"namespace_id": 1,
"schema_id": "asserted_payload",
"version": "v1"
}
Output:
{
"record": {
"created_at": {
"kind": "unix_millis",
"value": 1710000000000
},
"description": "Asserted payload schema.",
"namespace_id": 1,
"schema": {
"additionalProperties": false,
"properties": {
"deploy_env": {
"type": "string"
}
},
"required": [
"deploy_env"
],
"type": "object"
},
"schema_id": "asserted_payload",
"version": "v1"
}
}
scenarios_list
List registered scenarios for a namespace.
Inputs
cursor(optional, nullable): One of: null, string.limit(optional): Maximum number of records to return.namespace_id(required): Namespace identifier.
Outputs
items(required): Type: array.next_token(required, nullable): One of: null, string.
Notes
- Requires namespace_id.
- Returns scenario identifiers and hashes.
Example
List scenarios for a namespace.
Input:
{
"cursor": null,
"limit": 50,
"namespace_id": 1
}
Output:
{
"items": [
{
"namespace_id": 1,
"scenario_id": "example-scenario",
"scenario_law_identity": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
}
],
"next_token": null
}
decision_gate_docs_search
Search Decision Gate documentation for runtime guidance.
Inputs
max_sections(optional): Maximum number of sections to return (default 3, hard cap 10).query(required): Search query for documentation sections.
Outputs
docs_covered(required): Type: array.sections(required): Type: array.suggested_followups(required): Role-aware follow-up prompts.
Notes
- Use for quick lookups on typed evidence flow, predicates, and local acquisition.
- Returns ranked sections with role tags and suggested follow-ups.
- Search is deterministic and scoped to the configured doc catalog.
Example
Search for evidence admission and pure-evaluation guidance.
Input:
{
"max_sections": 2,
"query": "precheck versus live admitted evidence"
}
Output:
{
"docs_covered": [
{
"doc_id": "evidence_flow_and_execution_model",
"doc_role": "reasoning",
"doc_title": "Evidence Flow + Execution Model"
}
],
"sections": [
{
"content": "...",
"doc_id": "evidence_flow_and_execution_model",
"doc_role": "reasoning",
"doc_title": "Evidence Flow + Execution Model",
"heading": "Core Data Flow",
"rank": 0
}
],
"suggested_followups": [
"Refine the query with condition-domain or acquisition keywords for targeted guidance."
]
}