generated: '2026-08-13'
method: derived
source: mcp/coresignal-mcp.yml + openapi/_original/*.yml
note: >-
  Binds each Coresignal MCP tool to the OpenAPI operation(s) that back it. The v2 tool inputSchemas
  are OAuth-gated, so v2 rows are mapped by name and by the provider's own tool documentation and
  carry honest confidence values. The LEGACY server's three tools were introspected anonymously and
  their inputSchema (Elasticsearch Query DSL `query` + `keys` + `limit`) maps cleanly onto the
  search/es_dsl + collect operation pair, so those rows are high confidence.
surfaces:
  openapi:
    files:
      - openapi/_original/coresignal-multi-source-company-api-openapi.yml
      - openapi/_original/coresignal-multi-source-employee-api-openapi.yml
      - openapi/_original/coresignal-multi-source-jobs-api-openapi.yml
      - openapi/coresignal-search-api-openapi.yml
      - openapi/coresignal-collect-api-openapi.yml
    gated: false
  graphql:
    endpoint: null
    note: Coresignal publishes no GraphQL surface.
  mcp:
    url: https://mcp.coresignal.com/mcp/v2
    gated: true
    gate: OAuth 2.1 bearer required for tools/list
  mcp_legacy:
    url: https://mcp.coresignal.com/mcp
    gated: false
    introspected: '2026-08-13'
crosswalk:
  # --- legacy server (introspected, high confidence) ---
  - tool: coresignal_company_multisource_api
    server: legacy
    category: company
    rest: [searchCompaniesByEsDsl, collectCompany]
    binding: rest
    confidence: high
    note: >-
      inputSchema.query is an Elasticsearch Query DSL object matching the /search/es_dsl request body;
      the tool then collects the matched records, which is /collect/{id}. Composite of the two.
  - tool: coresignal_employee_multisource_api
    server: legacy
    category: employee
    rest: [searchEmployeesByEsDsl, collectEmployee]
    binding: rest
    confidence: high
    note: Same search-then-collect composite against the Multi-source Employee API.
  - tool: coresignal_job_api
    server: legacy
    category: job
    rest: [searchJobsByEsDsl, collectJob]
    binding: rest
    confidence: high
    note: >-
      Docs describe this tool as backed by the Base Jobs API; the catalog's OpenAPI covers the
      Multi-source Jobs API, whose search/collect shape is identical. Mapped to the operations we hold.
  # --- v2 server (gated schemas, mapped by provider documentation) ---
  - tool: entity_search
    server: v2
    category: search
    rest: [searchCompaniesByEsDsl, searchEmployeesByEsDsl, searchJobsByEsDsl, searchCompaniesByFilter, searchEmployeesByFilter, searchJobsByFilter]
    binding: rest
    confidence: medium
    note: >-
      Fans out across all three entities — the tool takes natural language, builds the Elasticsearch
      query server-side, and returns a preview. The preview behaviour corresponds to the documented
      /search/es_dsl/preview and /search/filter/preview endpoints, which are not in the OpenAPI we hold.
  - tool: entity_fetch
    server: v2
    category: collect
    rest: [collectCompany, collectEmployee, collectJob, bulkCollectCompanies, bulkCollectEmployees]
    binding: rest
    confidence: medium
    note: >-
      Up to 20 ids maps to per-id /collect/{id}; up to 1,000 records via a cache_id maps to the
      bulk-collect surface.
  - tool: email_enrich
    server: v2
    category: enrichment
    rest: []
    binding: none
    confidence: low
    note: >-
      Backed by the Contact Enrichment API, which is a separate paid endpoint not present in the
      OpenAPI files this repo holds. Listed here so the divergence is recorded, not to claim a binding.
mcp_only:
  - tool: entity_fields
    server: v2
    reason: >-
      Semantic search over an entity's ~300 field names. There is no public REST operation for field
      discovery — the equivalent for a REST consumer is reading the static Data Dictionary docs page.
  - tool: artifact_read
    server: v2
    reason: >-
      Pages rows out of a server-side result file the MCP server created. Pure MCP-server construct;
      the REST API has no artifact/file-delivery surface.
  - tool: email_enrich
    server: v2
    reason: >-
      Contact Enrichment API is on-demand/plan-gated and is not published in any OpenAPI Coresignal
      distributes, so there is no operationId to bind to.
rest_only:
  - capability: Structured filter search
    operations: [searchCompaniesByFilter, searchEmployeesByFilter, searchJobsByFilter]
    note: >-
      The v2 MCP surface takes natural language only; a REST consumer can post an explicit structured
      filter object. No dedicated tool exposes the filter schema.
  - capability: Bulk Collect
    operations: [bulkCollectCompanies, bulkCollectEmployees]
    note: >-
      Asynchronous bulk job submission (up to 10k ids, 30-day retrieval window) has no MCP tool;
      entity_fetch caps at 1,000 records per call.
  - capability: Webhook subscriptions
    operations: []
    note: >-
      Documented at /v2/subscriptions (create, list, renew, delete, simulate) but not present in the
      OpenAPI files this repo holds and not exposed as an MCP tool. Recorded in asyncapi/.
coverage:
  tools_named: 8
  tools_named_v2: 5
  tools_named_legacy: 3
  tools_bound_to_rest: 5
  mcp_only: 3
  rest_operations_total: 11
  rest_operations_with_a_tool: 11
  graphql_fields: 0