generated: '2026-08-09'
method: derived
source: >-
  mcp/avora-agent-forge-tools-list.json (live tools/list, 200) bound to
  openapi/avora-agent-forge-commerce-openapi.json and
  openapi/avora-agent-forge-public-sector-openapi.json
surfaces:
  openapi:
    - file: openapi/avora-agent-forge-commerce-openapi.json
      title: AVORA Agent Forge Commerce API
      version: 3.0.0
      operations: 5
    - file: openapi/avora-agent-forge-public-sector-openapi.json
      title: AVORA Public AI Interoperability API
      version: 1.0.0
      operations: 3
  mcp:
    url: https://avora-agent-forge.netlify.app/mcp
    transport: streamable-http
    gated: false
    tools: 11
  graphql: null
  a2a:
    url: https://avora-agent-forge.netlify.app/a2a
    skills: 3
    note: >-
      The A2A card publishes 3 skills that are a strict subset of the MCP tool set
      (list_avora_offers, scan_solana_token_public_sector, get_verified_revenue_status).
crosswalk:
  - tool: scan_solana_token
    category: risk-scan
    rest: [scanSolanaTokenFree]
    binding: rest
    confidence: high
    note: GET /api/agent/scan; tool `mint` maps to the required mint query parameter.
  - tool: scan_solana_token_public_sector
    category: risk-scan
    rest: [createSignedSolanaTokenAssessment]
    binding: rest
    confidence: high
    note: >-
      GET /api/gov/v1/scan on the public-sector spec; returns the Ed25519 JWS signed
      assessment. Tool adds an optional requestId trace parameter.
  - tool: list_avora_offers
    category: catalog
    rest: [listEvidenceOffers]
    binding: rest
    confidence: high
    note: GET /api/agent/offers; no required inputs on either surface.
  - tool: create_starter_snapshot_order
    category: order
    rest: [createPaidEvidenceOrder]
    binding: rest
    confidence: high
    note: >-
      Fans in to one REST operation. GET /api/agent/payment-order with tier=starter;
      the tier is fixed by the tool name rather than passed as a parameter.
  - tool: create_pro_report_order
    category: order
    rest: [createPaidEvidenceOrder]
    binding: rest
    confidence: high
    note: Same REST operation with tier=pro.
  - tool: create_portfolio_report_order
    category: order
    rest: [createPaidEvidenceOrder]
    binding: rest
    confidence: high
    note: >-
      Same REST operation with tier=portfolio; takes `mints` (2-10 comma-separated)
      instead of a single `mint`.
  - tool: verify_starter_snapshot_payment
    category: settlement
    rest: [verifyPaymentAndDeliverEvidence]
    binding: rest
    confidence: high
    note: >-
      POST /api/agent/pro-scan with tier=starter. Requires mint + reference + orderId,
      matching the PaymentProof schema.
  - tool: verify_pro_report_payment
    category: settlement
    rest: [verifyPaymentAndDeliverEvidence]
    binding: rest
    confidence: high
    note: Same REST operation with tier=pro.
  - tool: verify_portfolio_report_payment
    category: settlement
    rest: [verifyPaymentAndDeliverEvidence]
    binding: rest
    confidence: high
    note: Same REST operation with tier=portfolio, taking `mints`.
  - tool: get_verified_revenue_status
    category: telemetry
    rest: [getVerifiedRevenueStatus]
    binding: rest
    confidence: high
    note: GET /api/agent/revenue; optional `days` maps to the days query parameter (1-90).
mcp_only:
  - tool: get_avora_service_info
    reason: >-
      Service metadata tool with no corresponding public REST operation in either OpenAPI.
      The equivalent human/machine surface is the static /llms.txt and
      /.well-known/avora-agent.json descriptors rather than a REST endpoint.
rest_only:
  - operation: getPublicAiServiceHealth
    spec: openapi/avora-agent-forge-public-sector-openapi.json
    capability: health
    reason: Liveness probe (GET /api/gov/v1/health); not exposed as an agent tool.
  - operation: getReceiptVerificationKeys
    spec: openapi/avora-agent-forge-public-sector-openapi.json
    capability: key-discovery
    reason: >-
      GET /.well-known/jwks.json; receipt verification keys are fetched directly by the
      verifier, not through a tool call.
coverage:
  tools_named: 11
  tools_bound: 10
  mcp_only: 1
  rest_operations_total: 8
  rest_operations_with_tool: 6
  rest_only: 2
  notes: >-
    The MCP surface is wider than REST in tool count but narrower in capability: nine of
    the eleven tools collapse onto just three REST operations, because tier (starter / pro
    / portfolio) is encoded in the tool name instead of a parameter. That is a deliberate
    agent-ergonomics choice, not a capability divergence.