Decision Gate MCP Tools

Download: tooling.json (326 KB) Download: openapi/decision-gate.json (408 KB)

Tool detail pages

• scenario_define - Register a ScenarioSpec, validate it, and return the canonical hash used for integrity checks.
• scenario_start - Create a new run state for a scenario and optionally emit entry packets.
• scenario_status - Fetch a read-only run snapshot and safe summary without changing state.
• scenario_next - Evaluate gates in response to an agent-driven next request.
• scenario_submit - Submit external artifacts into run state for audit and later evaluation.
• scenario_trigger - Submit a trigger event (scheduler/external) and evaluate the run.
• evidence_query - Query an evidence provider with full run context and disclosure policy.
• runpack_export - Export deterministic runpack artifacts for offline verification.
• runpack_verify - Verify a runpack manifest and artifacts offline.
• providers_list - List registered evidence providers and capabilities summary.
• provider_contract_get - Fetch the canonical provider contract JSON and hash for a provider.
• provider_check_schema_get - Fetch check schema details (params/result/comparators) for a provider.
• schemas_register - Register a data shape schema for a tenant and namespace.
• schemas_list - List registered data shapes for a tenant and namespace.
• schemas_get - Fetch a specific data shape by identifier and version.
• scenarios_list - List registered scenarios for a tenant and namespace.
• precheck - Evaluate a scenario against asserted data without mutating state.
• decision_gate_docs_search - Search Decision Gate documentation for runtime guidance.
• typed_providers_import - Import OpenAPI into typed provider artifacts and register a profile version.
• typed_providers_register - Register prebuilt typed provider contract + runtime profile artifacts.
• typed_providers_list - List typed provider lifecycle catalogs and registered versions.
• typed_providers_get - Fetch typed provider artifacts + lifecycle metadata for one version.
• typed_providers_activate - Activate a typed provider lifecycle version.
• typed_providers_deprecate - Deprecate a typed provider lifecycle version, with optional active rollback.

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_define registers and validates a ScenarioSpec.
• scenario_start creates a run and optionally issues entry packets.
• scenario_next advances an agent-driven run; scenario_trigger advances time/external triggers.
• scenario_status polls run state without mutating it.
• scenario_submit appends external artifacts for audit and later checks.
• runpack_export and runpack_verify support offline verification.

Artifact references

• authoring.md: authoring formats and normalization guidance.
• [scenario.json](/downloads/decision-gate/examples/scenario.json): full ScenarioSpec example.
• examples/scenario.ron: authoring-friendly ScenarioSpec example.
• [run-config.json](/downloads/decision-gate/examples/run-config.json): run config example for scenario_start.
• [decision-gate.toml](/downloads/decision-gate/examples/decision-gate.toml): MCP config example for providers.

Tool	Description
scenario_define	Register a ScenarioSpec, validate it, and return the canonical hash used for integrity checks.
scenario_start	Create a new run state for a scenario and optionally emit entry packets.
scenario_status	Fetch a read-only run snapshot and safe summary without changing state.
scenario_next	Evaluate gates in response to an agent-driven next request.
scenario_submit	Submit external artifacts into run state for audit and later evaluation.
scenario_trigger	Submit a trigger event (scheduler/external) and evaluate the run.
evidence_query	Query an evidence provider with full run context and disclosure policy.
runpack_export	Export deterministic runpack artifacts for offline verification.
runpack_verify	Verify a runpack manifest and artifacts offline.
providers_list	List registered evidence providers and capabilities summary.
provider_contract_get	Fetch the canonical provider contract JSON and hash for a provider.
provider_check_schema_get	Fetch check schema details (params/result/comparators) for a provider.
schemas_register	Register a data shape schema for a tenant and namespace.
schemas_list	List registered data shapes for a tenant and namespace.
schemas_get	Fetch a specific data shape by identifier and version.
scenarios_list	List registered scenarios for a tenant and namespace.
precheck	Evaluate a scenario against asserted data without mutating state.
decision_gate_docs_search	Search Decision Gate documentation for runtime guidance.
typed_providers_import	Import OpenAPI into typed provider artifacts and register a profile version.
typed_providers_register	Register prebuilt typed provider contract + runtime profile artifacts.
typed_providers_list	List typed provider lifecycle catalogs and registered versions.
typed_providers_get	Fetch typed provider artifacts + lifecycle metadata for one version.
typed_providers_activate	Activate a typed provider lifecycle version.
typed_providers_deprecate	Deprecate a typed provider lifecycle version, with optional active rollback.

scenario_define

Register a ScenarioSpec, validate it, and return the canonical hash used for integrity checks.

Inputs

• spec (required): Scenario specification to register.

Outputs

• scenario_id (required): Scenario identifier.
• spec_hash (required): Type: object.

Notes

• Use before starting runs; scenario_id becomes the stable handle for later calls.
• Validates stage/gate/condition IDs, RET trees, and condition references.
• Spec hash is deterministic; store it for audit and runpack integrity.
• Fails closed on invalid specs or duplicate scenario IDs.

Example

Register the example scenario spec.

Input:

{
  "spec": {
    "conditions": [
      {
        "comparator": "equals",
        "condition_id": "env_is_prod",
        "expected": "production",
        "policy_tags": [],
        "query": {
          "check_id": "get",
          "params": {
            "key": "DEPLOY_ENV"
          },
          "provider_id": "env"
        }
      },
      {
        "comparator": "equals",
        "condition_id": "after_freeze",
        "expected": true,
        "policy_tags": [],
        "query": {
          "check_id": "after",
          "params": {
            "timestamp": 1710000000000
          },
          "provider_id": "time"
        }
      }
    ],
    "default_tenant_id": null,
    "namespace_id": 1,
    "policies": [],
    "scenario_id": "example-scenario",
    "schemas": [],
    "spec_version": "v1",
    "stages": [
      {
        "advance_to": {
          "kind": "terminal"
        },
        "entry_packets": [
          {
            "content_type": "application/json",
            "expiry": null,
            "packet_id": "packet-hello",
            "payload": {
              "kind": "json",
              "value": {
                "message": "hello",
                "purpose": "scenario entry packet"
              }
            },
            "policy_tags": [],
            "schema_id": "schema-hello",
            "visibility_labels": [
              "public"
            ]
          }
        ],
        "gates": [
          {
            "gate_id": "env_gate",
            "requirement": {
              "Condition": "env_is_prod"
            }
          },
          {
            "gate_id": "time_gate",
            "requirement": {
              "Condition": "after_freeze"
            }
          }
        ],
        "on_timeout": "fail",
        "stage_id": "main",
        "timeout": null
      }
    ]
  }
}

Output:

{
  "scenario_id": "example-scenario",
  "spec_hash": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  }
}

scenario_start

Create a new run state for a scenario and optionally emit entry packets.

Inputs

• issue_entry_packets (required): Issue entry packets immediately.
• run_config (required): Run configuration and dispatch targets.
• scenario_id (required): Scenario identifier.
• started_at (required): Caller-supplied run start timestamp.

Outputs

• current_stage_id (required): Current stage identifier.
• decisions (required): Type: array.
• dispatch_targets (required): Type: array.
• gate_evals (required): Type: array.
• namespace_id (required): Namespace identifier.
• packets (required): Type: array.
• run_id (required): Run identifier.
• scenario_id (required): Scenario identifier.
• spec_hash (required): Type: object.
• stage_entered_at (required): One of: object, object.
• status (required): Type: string.
• submissions (required): Type: array.
• tenant_id (required): Tenant identifier.
• tool_calls (required): Type: array.
• triggers (required): Type: array.

Notes

• Requires RunConfig (tenant_id, run_id, scenario_id, dispatch_targets).
• Use started_at to record the caller-supplied start timestamp.
• If issue_entry_packets is true, entry packets are disclosed immediately.
• Fails closed if run_id already exists or scenario_id is unknown.

Example

Start a run for the example scenario and issue entry packets.

Input:

{
  "issue_entry_packets": true,
  "run_config": {
    "dispatch_targets": [
      {
        "agent_id": "agent-alpha",
        "kind": "agent"
      }
    ],
    "namespace_id": 1,
    "policy_tags": [],
    "run_id": "run-0001",
    "scenario_id": "example-scenario",
    "tenant_id": 1
  },
  "scenario_id": "example-scenario",
  "started_at": {
    "kind": "unix_millis",
    "value": 1710000000000
  }
}

Output:

{
  "current_stage_id": "main",
  "decisions": [],
  "dispatch_targets": [
    {
      "agent_id": "agent-alpha",
      "kind": "agent"
    }
  ],
  "gate_evals": [],
  "namespace_id": 1,
  "packets": [],
  "run_id": "run-0001",
  "scenario_id": "example-scenario",
  "spec_hash": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "stage_entered_at": {
    "kind": "unix_millis",
    "value": 1710000000000
  },
  "status": "active",
  "submissions": [],
  "tenant_id": 1,
  "tool_calls": [],
  "triggers": []
}

scenario_status

Fetch a read-only run snapshot and safe summary without changing state.

Inputs

• request (required): Status request payload.
• scenario_id (required): Scenario identifier.

Outputs

• current_stage_id (required): Current stage identifier.
• issued_packet_ids (required): Type: array.
• last_decision (required, nullable): One of: null, object.
• namespace_id (optional): Namespace identifier.
• run_id (required): Run identifier.
• safe_summary (required, nullable): One of: null, object.
• scenario_id (required): Scenario identifier.
• status (required): Type: string.

Notes

• Use for polling or UI state; does not evaluate gates.
• Safe summaries omit evidence values and may include retry hints.
• Returns issued packet IDs to help track disclosures.

Example

Poll run status without advancing the run.

Input:

{
  "request": {
    "correlation_id": null,
    "namespace_id": 1,
    "requested_at": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "run_id": "run-0001",
    "tenant_id": 1
  },
  "scenario_id": "example-scenario"
}

Output:

{
  "current_stage_id": "main",
  "issued_packet_ids": [],
  "last_decision": null,
  "run_id": "run-0001",
  "safe_summary": null,
  "scenario_id": "example-scenario",
  "status": "active"
}

scenario_next

Evaluate gates in response to an agent-driven next request.

Inputs

• feedback (optional, nullable): Optional feedback level override for scenario_next.
• request (required): Next request payload from an agent.
• scenario_id (required): Scenario identifier.

Outputs

• decision (required): Type: object.
• feedback (optional, nullable): One of: null, object.
• packets (required): Type: array.
• status (required): Type: string.

Notes

• Idempotent by trigger_id; repeated calls return the same decision.
• Records decision, evidence, and packet disclosures in run state.
• Requires an active run; completed or failed runs do not advance.
• Optional feedback can include gate trace or evidence when permitted by server feedback policy.

Example

Example 1: Evaluate the next agent-driven step for a run.

Input:

{
  "request": {
    "agent_id": "agent-alpha",
    "correlation_id": null,
    "namespace_id": 1,
    "run_id": "run-0001",
    "tenant_id": 1,
    "time": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "trigger_id": "trigger-0001"
  },
  "scenario_id": "example-scenario"
}

Output:

{
  "decision": {
    "correlation_id": null,
    "decided_at": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "decision_id": "decision-0001",
    "outcome": {
      "kind": "complete",
      "stage_id": "main"
    },
    "seq": 0,
    "stage_id": "main",
    "trigger_id": "trigger-0001"
  },
  "packets": [],
  "status": "completed"
}

Example 2: Evaluate a run and request trace feedback.

Input:

{
  "feedback": "trace",
  "request": {
    "agent_id": "agent-alpha",
    "correlation_id": null,
    "namespace_id": 1,
    "run_id": "run-0001",
    "tenant_id": 1,
    "time": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "trigger_id": "trigger-0001"
  },
  "scenario_id": "example-scenario"
}

Output:

{
  "decision": {
    "correlation_id": null,
    "decided_at": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "decision_id": "decision-0001",
    "outcome": {
      "kind": "complete",
      "stage_id": "main"
    },
    "seq": 0,
    "stage_id": "main",
    "trigger_id": "trigger-0001"
  },
  "feedback": {
    "gate_evaluations": [],
    "level": "trace"
  },
  "packets": [],
  "status": "completed"
}

scenario_submit

Submit external artifacts into run state for audit and later evaluation.

Inputs

• request (required): Submission payload and metadata.
• scenario_id (required): Scenario identifier.

Outputs

• record (required): Type: object.

Notes

• Payload is hashed and stored as a submission record.
• Payload is persisted in run state/runpack logs; do not send raw secrets.
• Does not advance the run by itself.
• Use for artifacts the model or operator supplies.

Example

Submit an external artifact for audit and later evaluation.

Input:

{
  "request": {
    "content_type": "application/json",
    "correlation_id": null,
    "namespace_id": 1,
    "payload": {
      "kind": "json",
      "value": {
        "artifact": "attestation",
        "status": "approved"
      }
    },
    "run_id": "run-0001",
    "submission_id": "submission-0001",
    "submitted_at": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "tenant_id": 1
  },
  "scenario_id": "example-scenario"
}

Output:

{
  "record": {
    "content_hash": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    },
    "content_type": "application/json",
    "correlation_id": null,
    "payload": {
      "kind": "json",
      "value": {
        "artifact": "attestation",
        "status": "approved"
      }
    },
    "run_id": "run-0001",
    "submission_id": "submission-0001",
    "submitted_at": {
      "kind": "unix_millis",
      "value": 1710000000000
    }
  }
}

scenario_trigger

Submit a trigger event (scheduler/external) and evaluate the run.

Inputs

• scenario_id (required): Scenario identifier.
• trigger (required): Trigger event payload.

Outputs

• decision (required): Type: object.
• packets (required): Type: array.
• status (required): Type: string.

Notes

• Trigger time is supplied by the caller; no wall-clock reads.
• Records the trigger event and resulting decision.
• Payload is persisted in run state/runpack logs; do not send raw secrets.
• Use for time-based or external system triggers.

Example

Advance a run from a scheduler or external trigger.

Input:

{
  "scenario_id": "example-scenario",
  "trigger": {
    "correlation_id": null,
    "kind": "tick",
    "namespace_id": 1,
    "payload": null,
    "run_id": "run-0001",
    "source_id": "scheduler-01",
    "tenant_id": 1,
    "time": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "trigger_id": "trigger-0001"
  }
}

Output:

{
  "decision": {
    "correlation_id": null,
    "decided_at": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "decision_id": "decision-0001",
    "outcome": {
      "kind": "complete",
      "stage_id": "main"
    },
    "seq": 0,
    "stage_id": "main",
    "trigger_id": "trigger-0001"
  },
  "packets": [],
  "status": "completed"
}

evidence_query

Query an evidence provider with full run context and disclosure policy.

Inputs

• context (required): Evidence context used for evaluation.
• query (required): Evidence query payload.

Outputs

• result (required): Type: object.

Notes

• Disclosure policy may redact raw values; hashes/anchors still returned.
• Use for diagnostics or preflight checks; runtime uses the same provider logic.
• Requires provider_id, check_id, and full EvidenceContext.

Example

Query an evidence provider using the run context.

Input:

{
  "context": {
    "correlation_id": null,
    "namespace_id": 1,
    "run_id": "run-0001",
    "scenario_id": "example-scenario",
    "stage_id": "main",
    "tenant_id": 1,
    "trigger_id": "trigger-0001",
    "trigger_time": {
      "kind": "unix_millis",
      "value": 1710000000000
    }
  },
  "query": {
    "check_id": "get",
    "params": {
      "key": "DEPLOY_ENV"
    },
    "provider_id": "env"
  }
}

Output:

{
  "result": {
    "content_type": "text/plain",
    "error": null,
    "evidence_anchor": {
      "anchor_type": "env",
      "anchor_value": "DEPLOY_ENV"
    },
    "evidence_hash": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    },
    "evidence_ref": null,
    "lane": "verified",
    "signature": null,
    "value": {
      "kind": "json",
      "value": "production"
    }
  }
}

runpack_export

Export deterministic runpack artifacts for offline verification.

Inputs

• generated_at (required): Timestamp recorded in the manifest.
• include_verification (required): Generate a verification report artifact.
• manifest_name (optional, nullable): Optional override for the manifest file name.
• namespace_id (required): Namespace identifier.
• output_dir (optional, nullable): Optional output directory (required for filesystem export).
• run_id (required): Run identifier.
• scenario_id (required): Scenario identifier.
• tenant_id (required): Tenant identifier.

Outputs

• manifest (required): Type: object.
• report (required, nullable): One of: null, object.
• storage_uri (optional, nullable): Optional storage URI for managed runpack storage backends.

Notes

• Writes manifest and logs to output_dir; generated_at is recorded in the manifest.
• include_verification adds a verification report artifact.
• Export-time report.checked_files excludes verifier_report.json; offline runpack_verify checked_files includes it (+1 for the same runpack).
• Use after runs complete or for audit snapshots.

Example

Export a runpack with manifest metadata.

Input:

{
  "generated_at": {
    "kind": "unix_millis",
    "value": 1710000000000
  },
  "include_verification": false,
  "manifest_name": "manifest.json",
  "namespace_id": 1,
  "output_dir": "/var/lib/decision-gate/runpacks/run-0001",
  "run_id": "run-0001",
  "scenario_id": "example-scenario",
  "tenant_id": 1
}

Output:

{
  "manifest": {
    "artifacts": [
      {
        "artifact_id": "decision_log",
        "content_type": "application/json",
        "hash": {
          "algorithm": "sha256",
          "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
        },
        "kind": "decision_log",
        "path": "decision_log.json",
        "required": true
      }
    ],
    "generated_at": {
      "kind": "unix_millis",
      "value": 1710000000000
    },
    "hash_algorithm": "sha256",
    "integrity": {
      "file_hashes": [
        {
          "hash": {
            "algorithm": "sha256",
            "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
          },
          "path": "decision_log.json"
        }
      ],
      "root_hash": {
        "algorithm": "sha256",
        "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
      }
    },
    "manifest_version": "v1",
    "namespace_id": 1,
    "run_id": "run-0001",
    "scenario_id": "example-scenario",
    "spec_hash": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    },
    "tenant_id": 1,
    "verifier_mode": "offline_strict"
  },
  "report": null,
  "storage_uri": null
}

runpack_verify

Verify a runpack manifest and artifacts offline.

Inputs

• manifest_path (required): Manifest path relative to runpack root.
• runpack_dir (required): Runpack root directory.

Outputs

• report (required): Type: object.
• status (required): Runpack verification status.

Notes

• Validates hashes, integrity root, and decision log structure.
• Fails closed on missing or tampered files.
• Use in CI or offline audit pipelines.

Example

Verify a runpack manifest and artifacts offline.

Input:

{
  "manifest_path": "manifest.json",
  "runpack_dir": "/var/lib/decision-gate/runpacks/run-0001"
}

Output:

{
  "report": {
    "checked_files": 12,
    "errors": [],
    "status": "pass"
  },
  "status": "pass"
}

providers_list

List registered evidence providers and capabilities summary.

Inputs

Outputs

• providers (required): Type: array.

Notes

• Returns provider identifiers and transport metadata.
• Results are scoped by auth policy.

Example

List registered evidence providers.

Input:

{}

Output:

{
  "providers": [
    {
      "checks": [
        "get"
      ],
      "provider_id": "env",
      "transport": "builtin"
    }
  ]
}

provider_contract_get

Fetch the canonical provider contract JSON and hash for a provider.

Inputs

• provider_id (required): Provider identifier.

Outputs

• contract (required): Type: object.
• contract_hash (required): Type: object.
• provider_id (required): Provider identifier.
• source (required): Contract source origin.
• version (required, nullable): Optional contract version label.

Notes

• Returns the provider contract as loaded by the MCP server.
• Includes a canonical hash for audit and reproducibility.
• Subject to provider disclosure policy and authz.

Example

Fetch the contract JSON for a provider.

Input:

{
  "provider_id": "json"
}

Output:

{
  "contract": {
    "checks": [],
    "config_schema": {
      "additionalProperties": false,
      "type": "object"
    },
    "description": "Reads JSON or YAML files and evaluates JSONPath.",
    "name": "JSON Provider",
    "notes": [],
    "provider_id": "json",
    "transport": "builtin"
  },
  "contract_hash": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "provider_id": "json",
  "source": "builtin",
  "version": null
}

provider_check_schema_get

Fetch check schema details (params/result/comparators) for a provider.

Inputs

• check_id (required): Provider check identifier.
• provider_id (required): Provider identifier.

Outputs

• allowed_comparators (required): Comparator allow-list for this check.
• anchor_types (required): Anchor types emitted by this check.
• check_id (required): Check identifier.
• content_types (required): Content types for check output.
• contract_hash (required): Type: object.
• determinism (required): Determinism classification for provider checks.
• examples (required): Type: array.
• params_required (required): Whether params are required for this check.
• params_schema (required): JSON schema for check params.
• provider_id (required): Provider identifier.
• result_schema (required): JSON schema for check result value.

Notes

• Returns compiled schema metadata for a single check.
• Includes comparator allow-lists and check examples.
• Subject to provider disclosure policy and authz.

Example

Fetch check schema details for a provider.

Input:

{
  "check_id": "path",
  "provider_id": "json"
}

Output:

{
  "allowed_comparators": [
    "equals",
    "in_set",
    "exists",
    "not_exists"
  ],
  "anchor_types": [],
  "check_id": "path",
  "content_types": [
    "application/json"
  ],
  "contract_hash": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "determinism": "external",
  "examples": [],
  "params_required": true,
  "params_schema": {
    "properties": {
      "file": {
        "type": "string"
      },
      "jsonpath": {
        "type": "string"
      }
    },
    "required": [
      "file"
    ],
    "type": "object"
  },
  "provider_id": "json",
  "result_schema": {
    "type": [
      "null",
      "string",
      "number",
      "boolean",
      "array",
      "object"
    ]
  }
}

schemas_register

Register a data shape schema for a tenant and 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",
    "tenant_id": 1,
    "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",
    "tenant_id": 1,
    "version": "v1"
  }
}

schemas_list

List registered data shapes for a tenant and namespace.

Inputs

• cursor (optional, nullable): One of: null, string.
• limit (optional): Maximum number of records to return.
• namespace_id (required): Namespace identifier.
• tenant_id (required): Tenant identifier.

Outputs

• items (required): Type: array.
• next_token (required, nullable): One of: null, string.

Notes

• Requires tenant_id and namespace_id.
• Supports pagination via cursor + limit.

Example

List data shapes for a namespace.

Input:

{
  "cursor": null,
  "limit": 50,
  "namespace_id": 1,
  "tenant_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",
      "tenant_id": 1,
      "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.
• tenant_id (required): Tenant identifier.
• version (required): Data shape version identifier.

Outputs

• record (required): Type: object.

Notes

• Requires tenant_id, 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",
  "tenant_id": 1,
  "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",
    "tenant_id": 1,
    "version": "v1"
  }
}

scenarios_list

List registered scenarios for a tenant and namespace.

Inputs

• cursor (optional, nullable): One of: null, string.
• limit (optional): Maximum number of records to return.
• namespace_id (required): Namespace identifier.
• tenant_id (required): Tenant identifier.

Outputs

• items (required): Type: array.
• next_token (required, nullable): One of: null, string.

Notes

• Requires tenant_id and namespace_id.
• Returns scenario identifiers and hashes.

Example

List scenarios for a namespace.

Input:

{
  "cursor": null,
  "limit": 50,
  "namespace_id": 1,
  "tenant_id": 1
}

Output:

{
  "items": [
    {
      "namespace_id": 1,
      "scenario_id": "example-scenario",
      "spec_hash": {
        "algorithm": "sha256",
        "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
      }
    }
  ],
  "next_token": null
}

precheck

Evaluate a scenario against asserted data without mutating state.

Inputs

• data_shape (required): Type: object.
• namespace_id (required): Namespace identifier.
• payload (required): Asserted data payload.
• scenario_id (optional, nullable): One of: null, string.
• spec (optional, nullable): One of: null, ref decision-gate://contract/schemas/scenario.schema.json.
• stage_id (optional, nullable): One of: null, string.
• tenant_id (required): Tenant identifier.

Outputs

• decision (required): One of: object, object, object, object, object.
• gate_evaluations (required): Type: array.

Notes

• Validates asserted data against a registered shape.
• Does not mutate run state; intended for simulation.

Example

Precheck a scenario with asserted data.

Input:

{
  "data_shape": {
    "schema_id": "asserted_payload",
    "version": "v1"
  },
  "namespace_id": 1,
  "payload": {
    "deploy_env": "production"
  },
  "scenario_id": "example-scenario",
  "spec": null,
  "stage_id": null,
  "tenant_id": 1
}

Output:

{
  "decision": {
    "kind": "hold",
    "summary": {
      "policy_tags": [],
      "retry_hint": "await_evidence",
      "status": "hold",
      "unmet_gates": [
        "ready"
      ]
    }
  },
  "gate_evaluations": []
}

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 evidence flow, comparators, and provider semantics.
• Returns ranked sections with role tags and suggested follow-ups.
• Search is deterministic and scoped to the configured doc catalog.

Example

Search for evidence flow and trust lane guidance.

Input:

{
  "max_sections": 2,
  "query": "precheck vs live run trust lanes"
}

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 comparator or provider keywords for targeted guidance."
  ]
}

typed_providers_import

Import OpenAPI into typed provider artifacts and register a profile version.

Inputs

• activate (optional): Activate this version after registration.
• allow_unsafe_methods (optional): Allow non-safe HTTP methods during import.
• credential_bindings (required): OpenAPI security scheme id to structured credential binding metadata (locator + value_render). Use {} for unauthenticated imports.
• external_ref_mode (optional): Optional external $ref resolution policy.
• max_response_bytes (optional): Optional runtime response-size limit override.
• media_support_mode (optional): Optional media support mode.
• namespace_id (required): Namespace identifier.
• openapi (required): OpenAPI 3.x document as JSON.
• openapi_conformance_mode (optional): Conformance mode: strict blocks unsupported constructs; audit reports without activation.
• openapi_semantics_mode (optional): Optional OpenAPI semantics mode override.
• openapi_source_path (optional): Optional source path used to resolve local-file refs.
• operation_allowlist (required): Explicit operationId allowlist for strict import.
• outbound_max_inflight (optional): Optional outbound inflight limit override.
• provider_id (required): Typed provider identifier.
• retry_initial_backoff_ms (optional): Optional outbound retry initial-backoff override.
• retry_max_attempts (optional): Optional outbound retry max-attempts override.
• retry_max_backoff_ms (optional): Optional outbound retry max-backoff override.
• tenant_id (required): Tenant identifier.
• timeout_ms (optional): Optional runtime timeout override in milliseconds.
• version (required): Lifecycle version identifier.

Outputs

• active_version (required, nullable): One of: null, string.
• conformance_summary (required): Type: object.
• contract_hash (required): Type: object.
• operation_count (required): Type: integer.
• profile_digest (required): Type: object.
• provider_id (required): Typed provider identifier.
• register_outcome (required): Typed profile lifecycle registration outcome.
• source_digest (required): Type: object.
• version (required): Lifecycle version identifier.

Notes

• Requires tenant_id and namespace_id scope fields.
• Strict fail-closed import requires explicit operation allowlist.
• Requires explicit credential_bindings mapping; use {} for unauthenticated imports.
• Supports bounded outbound limits and retry policy for generated runtime profiles.
• Generates deterministic provider contract + runtime profile artifacts.
• Registers lifecycle metadata and optionally activates the imported version.

Example

Import an OpenAPI document into typed artifacts.

Input:

{
  "activate": true,
  "allow_unsafe_methods": false,
  "credential_bindings": {},
  "external_ref_mode": "local_file_only",
  "max_response_bytes": 1048576,
  "media_support_mode": "json_only",
  "namespace_id": 1,
  "openapi": {
    "openapi": "3.1.0",
    "paths": {
      "/assets": {
        "get": {
          "operationId": "listAssets",
          "responses": {
            "200": {
              "content": {
                "application/json": {
                  "schema": {
                    "additionalProperties": false,
                    "properties": {
                      "items": {
                        "type": "array"
                      }
                    },
                    "required": [
                      "items"
                    ],
                    "type": "object"
                  }
                }
              },
              "description": "ok"
            }
          }
        }
      }
    }
  },
  "openapi_conformance_mode": "strict",
  "openapi_semantics_mode": "auto",
  "operation_allowlist": [
    "listAssets"
  ],
  "outbound_max_inflight": 32,
  "provider_id": "asset_api",
  "retry_initial_backoff_ms": 100,
  "retry_max_attempts": 3,
  "retry_max_backoff_ms": 2000,
  "tenant_id": 1,
  "timeout_ms": 5000,
  "version": "2026-02-17.1"
}

Output:

{
  "active_version": "2026-02-17.1",
  "conformance_summary": {
    "activation_allowed": true,
    "import_mode": "strict",
    "supported_feature_hits": [
      "openapi.version.3_1"
    ],
    "supported_features_covered": 1,
    "unsupported_codes": [],
    "unsupported_construct_count": 0,
    "unsupported_findings": []
  },
  "contract_hash": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "operation_count": 1,
  "profile_digest": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "provider_id": "asset_api",
  "register_outcome": "registered",
  "source_digest": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "version": "2026-02-17.1"
}

typed_providers_register

Register prebuilt typed provider contract + runtime profile artifacts.

Inputs

• activate (optional): Activate this version after registration.
• contract (required): Type: object.
• namespace_id (required): Namespace identifier.
• provider_id (required): Typed provider identifier.
• runtime_profile (required): Typed runtime profile artifact.
• tenant_id (required): Tenant identifier.
• version (required): Lifecycle version identifier.

Outputs

• active_version (required, nullable): One of: null, string.
• contract_hash (required): Type: object.
• operation_count (required): Type: integer.
• profile_digest (required): Type: object.
• provider_id (required): Typed provider identifier.
• register_outcome (required): Typed profile lifecycle registration outcome.
• source_digest (required): Type: object.
• version (required): Lifecycle version identifier.

Notes

• Requires tenant_id and namespace_id scope fields.
• Validates typed transport/provider identity consistency.
• Validates runtime profile digest integrity before registration.
• Registers lifecycle metadata and optionally activates the version.

Example

Register prebuilt typed provider artifacts.

Input:

{
  "activate": false,
  "contract": {
    "checks": [],
    "config_schema": {
      "additionalProperties": false,
      "type": "object"
    },
    "description": "Typed API provider",
    "name": "Asset API",
    "notes": [],
    "provider_id": "asset_api",
    "transport": "typed"
  },
  "namespace_id": 1,
  "provider_id": "asset_api",
  "runtime_profile": {
    "anchors": {
      "anchor_type": "typed_request"
    },
    "operations": [
      {
        "allowed_status_codes": [
          200
        ],
        "auth": {
          "alternatives": [
            {
              "kind": "none"
            }
          ]
        },
        "bindings": [],
        "check_id": "listAssets",
        "content_types": [
          "application/json"
        ],
        "extraction": {
          "kind": "json_body"
        },
        "method": "GET",
        "operation_id": "listAssets",
        "target": "https://api.example.com/assets"
      }
    ],
    "profile_digest": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    },
    "profile_version": "typed-runtime-v4",
    "protocol": "openapi_http",
    "provider_id": "asset_api",
    "security": {
      "allow_http": false,
      "allow_private_networks": false,
      "allowed_hosts": null,
      "max_response_bytes": 1048576,
      "outbound_max_inflight": 32,
      "retry": {
        "initial_backoff_ms": 100,
        "max_attempts": 3,
        "max_backoff_ms": 2000
      },
      "timeout_ms": 5000
    },
    "source_digest": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    }
  },
  "tenant_id": 1,
  "version": "2026-02-17.2"
}

Output:

{
  "active_version": null,
  "contract_hash": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "operation_count": 1,
  "profile_digest": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "provider_id": "asset_api",
  "register_outcome": "registered",
  "source_digest": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "version": "2026-02-17.2"
}

typed_providers_list

List typed provider lifecycle catalogs and registered versions.

Inputs

• cursor (optional, nullable): One of: null, string.
• limit (optional): Maximum number of provider catalogs to return.
• namespace_id (required): Namespace identifier.
• tenant_id (required): Tenant identifier.

Outputs

• items (required): Type: array.
• next_token (required, nullable): One of: null, string.

Notes

• Requires tenant_id and namespace_id scope fields.
• Supports deterministic pagination via cursor + limit.
• Results are scoped by provider discovery disclosure policy.

Example

List typed provider lifecycle catalogs.

Input:

{
  "cursor": null,
  "limit": 50,
  "namespace_id": 1,
  "tenant_id": 1
}

Output:

{
  "items": [
    {
      "active_version": "2026-02-17.1",
      "provider_id": "asset_api",
      "versions": [
        {
          "deprecated": false,
          "profile_digest": {
            "algorithm": "sha256",
            "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
          },
          "source_digest": {
            "algorithm": "sha256",
            "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
          },
          "version": "2026-02-17.1"
        }
      ]
    }
  ],
  "next_token": null
}

typed_providers_get

Fetch typed provider artifacts + lifecycle metadata for one version.

Inputs

• namespace_id (required): Namespace identifier.
• observed_profile_digest (optional, nullable): Optional profile digest observation for drift checks.
• observed_source_digest (optional, nullable): Optional source digest observation for drift checks.
• provider_id (required): Typed provider identifier.
• tenant_id (required): Tenant identifier.
• version (optional, nullable): One of: null, string.

Outputs

• active_version (required, nullable): One of: null, string.
• catalog_version (required): Typed lifecycle catalog schema version.
• contract (required): Type: object.
• drift_status (required, nullable): One of: null, string.
• provider_id (required): Typed provider identifier.
• record (required): Type: object.
• runtime_profile (required): Typed runtime profile artifact.
• selected_version (required): Selected lifecycle version.
• source (required): How the typed artifacts were registered.

Notes

• Requires tenant_id and namespace_id scope fields.
• Returns provider contract + runtime profile for the selected version.
• Can evaluate digest drift against observed source/profile digests.
• If version is omitted, active version is selected.

Example

Fetch typed provider artifacts and lifecycle metadata.

Input:

{
  "namespace_id": 1,
  "observed_profile_digest": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "observed_source_digest": {
    "algorithm": "sha256",
    "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
  },
  "provider_id": "asset_api",
  "tenant_id": 1,
  "version": "2026-02-17.1"
}

Output:

{
  "active_version": "2026-02-17.1",
  "catalog_version": "typed-lifecycle-v1",
  "contract": {
    "checks": [],
    "config_schema": {
      "additionalProperties": false,
      "type": "object"
    },
    "description": "Typed API provider",
    "name": "Asset API",
    "notes": [],
    "provider_id": "asset_api",
    "transport": "typed"
  },
  "drift_status": "in_sync",
  "provider_id": "asset_api",
  "record": {
    "deprecated": false,
    "profile_digest": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    },
    "source_digest": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    },
    "version": "2026-02-17.1"
  },
  "runtime_profile": {
    "anchors": {
      "anchor_type": "typed_request"
    },
    "operations": [
      {
        "allowed_status_codes": [
          200
        ],
        "auth": {
          "alternatives": [
            {
              "kind": "none"
            }
          ]
        },
        "bindings": [],
        "check_id": "listAssets",
        "content_types": [
          "application/json"
        ],
        "extraction": {
          "kind": "json_body"
        },
        "method": "GET",
        "operation_id": "listAssets",
        "target": "https://api.example.com/assets"
      }
    ],
    "profile_digest": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    },
    "profile_version": "typed-runtime-v4",
    "protocol": "openapi_http",
    "provider_id": "asset_api",
    "security": {
      "allow_http": false,
      "allow_private_networks": false,
      "allowed_hosts": null,
      "max_response_bytes": 1048576,
      "outbound_max_inflight": 32,
      "retry": {
        "initial_backoff_ms": 100,
        "max_attempts": 3,
        "max_backoff_ms": 2000
      },
      "timeout_ms": 5000
    },
    "source_digest": {
      "algorithm": "sha256",
      "value": "5c3a5b6bce0f4a2c9e22c4fa6a1e6d8d90b0f2dfed1b7f1e9b3d3b3d1f0c9b21"
    }
  },
  "selected_version": "2026-02-17.1",
  "source": "openapi_import"
}

typed_providers_activate

Activate a typed provider lifecycle version.

Inputs

• namespace_id (required): Namespace identifier.
• provider_id (required): Typed provider identifier.
• tenant_id (required): Tenant identifier.
• version (required): Lifecycle version to activate.

Outputs

• active_version (required): Activated version.
• previous_active_version (required, nullable): One of: null, string.
• provider_id (required): Typed provider identifier.

Notes

• Requires tenant_id and namespace_id scope fields.
• Activation records prior active versions for rollback.
• Deprecated versions cannot be activated.

Example

Activate a typed provider version.

Input:

{
  "namespace_id": 1,
  "provider_id": "asset_api",
  "tenant_id": 1,
  "version": "2026-02-17.2"
}

Output:

{
  "active_version": "2026-02-17.2",
  "previous_active_version": "2026-02-17.1",
  "provider_id": "asset_api"
}

typed_providers_deprecate

Deprecate a typed provider lifecycle version, with optional active rollback.

Inputs

• namespace_id (required): Namespace identifier.
• provider_id (required): Typed provider identifier.
• rollback_if_active (optional): Rollback active version before deprecating when required.
• tenant_id (required): Tenant identifier.
• version (required): Lifecycle version to deprecate.

Outputs

• active_version (required, nullable): One of: null, string.
• deprecated_version (required): Deprecated lifecycle version.
• provider_id (required): Typed provider identifier.
• rolled_back_from (required, nullable): One of: null, string.

Notes

• Requires tenant_id and namespace_id scope fields.
• Deprecating the active version requires explicit rollback_if_active=true.
• Rollback selects the most recent non-deprecated prior active version.

Example

Deprecate a typed provider version with rollback.

Input:

{
  "namespace_id": 1,
  "provider_id": "asset_api",
  "rollback_if_active": true,
  "tenant_id": 1,
  "version": "2026-02-17.2"
}

Output:

{
  "active_version": "2026-02-17.1",
  "deprecated_version": "2026-02-17.2",
  "provider_id": "asset_api",
  "rolled_back_from": "2026-02-17.2"
}