generated: '2026-08-06'
method: derived
source: >-
  Derived by binding the candidate MCP tool list in mcp/apriori-mcp.yml to the operations in
  openapi/apriori-ap-connect-agent.yml (12 operations, transcribed from aPriori's published aP Connect
  Agent REST API Reference Guide).
purpose: >-
  Make the tool the first-class unit and bind it to the REST operation that backs it, so a tool's real
  inputSchema is the operation's parameters + requestBody rather than a guess. For aPriori this is a
  one-to-one mapping: there is no MCP server and no GraphQL surface, so every tool is REST-backed and no
  surface is a superset of another.
caveat: >-
  aPriori ships NO MCP server. The tool side of this crosswalk is API Evangelist's candidate derivation
  (mcp/apriori-mcp.yml, status: candidate). The REST side is real. Read this as "what a tool layer over this
  API would bind to", not as a description of a vendor product.
surfaces:
  rest_openapi: openapi/apriori-ap-connect-agent.yml
  rest_openapi_status: >-
    Transcribed from aPriori's published reference guide. The vendor's own machine-readable spec is served
    by the Agent at http://localhost:<port_number>/v4/api-docs — gated behind a customer deployment, not
    fetchable from any public host.
  graphql: null
  mcp: null
  mcp_status: does not exist
crosswalk:
- tool: get_service_configuration
  category: agent
  rest: [getServiceConfiguration]
  binding: rest
  confidence: high
- tool: get_service_status
  category: agent
  rest: [getServiceStatus]
  binding: rest
  confidence: high
- tool: create_shutdown_nonce
  category: agent
  rest: [createShutdownNonce]
  binding: rest
  confidence: high
- tool: initiate_shutdown
  category: agent
  rest: [initiateShutdown]
  binding: rest
  confidence: medium
  note: >-
    The backing operation's path carries an inferred {nonce} segment — aPriori's reference page renders the
    request line as POST /api/shutdown while documenting nonce as a required PATH parameter. Confidence is
    medium for that reason, not because the operation is uncertain.
- tool: list_workflows
  category: workflow
  rest: [listWorkflows]
  binding: rest
  confidence: high
- tool: get_workflow
  category: workflow
  rest: [getWorkflow]
  binding: rest
  confidence: high
- tool: run_workflow
  category: workflow
  rest: [runWorkflowAction]
  binding: rest
  confidence: high
  note: >-
    One operation, two behaviours selected by the {action} path parameter — run and runPartList. A tool
    layer would most usefully split these into two tools with different inputSchemas, since only runPartList
    takes a body.
- tool: list_workflow_jobs
  category: workflow
  rest: [listWorkflowJobs]
  binding: rest
  confidence: high
- tool: get_workflow_job
  category: workflow
  rest: [getWorkflowJob]
  binding: rest
  confidence: high
- tool: cancel_workflow_job
  category: workflow
  rest: [runWorkflowJobAction]
  binding: rest
  confidence: high
  note: 'Published action vocabulary is [cancel] only.'
- tool: get_job_results
  category: results
  rest: [getJobResults]
  binding: rest
  confidence: high
- tool: get_part_results
  category: results
  rest: [getPartResults]
  binding: rest
  confidence: high
mcp_only: []
rest_only: []
coverage:
  rest_operations: 12
  tools: 12
  bound: 12
  mcp_only: 0
  rest_only: 0
  bound_pct: 100
  note: >-
    100% because the tool list was derived FROM the operations. This is a completeness statement about the
    derivation, not evidence of a vendor tool surface.