generated: '2026-09-19'
method: derived
source: mcp/webcrawlerapi-com-mcp.yml (tool + inputSchema from src/index.ts) x openapi/webcrawlerapi-com-openapi.yml
  (24 operations; operationIds assigned by overlays/webcrawlerapi-com-openapi-overlay.yaml because the provider
  ships none)
surfaces:
  openapi:
    path: openapi/webcrawlerapi-com-openapi.yml
    original: openapi/_original/webcrawlerapi-com-swagger.json
    operations: 24
    gated: false
  mcp:
    mode: local-stdio
    package: webcrawler-mcp
    tools: 1
    gated: false
    note: Schema read from source, not from a live tools/list - there is no remote endpoint to call.
  graphql: null
  a2a:
    card: a2a/webcrawlerapi-com-agent-card.json
    note: Card declares 3 skills (scrape, crawl, crawling-agent) but no A2A endpoint answers; skills map to REST
      as noted in a2a/.
operation_id_note: operationIds below are API Evangelist overlay names; the method+path is the provider's own identifier.
crosswalk:
- tool: webcrawler-scrape
  category: scrape
  rest:
  - createScrapeV2
  rest_paths:
  - POST /v2/scrape
  binding: direct
  confidence: high
  note: 'src/index.ts POSTs {url, prompt?, main_content_only?} to https://api.webcrawlerapi.com/v2/scrape and returns
    markdown/page_title/page_status_code (or structured_data when a prompt is set). The tool exposes 3 of the 9
    PostScrapeRequestV2 fields: output_formats, output_format, clean_selectors, respect_robots_txt, max_age and
    response_schema are not reachable through MCP. The tool inherits the operation''s real inputSchema for the fields
    it does expose.'
mcp_only: []
rest_only:
- operationId: ping
  path: GET /ping
- operationId: runAgent
  path: POST /v1/agent
- operationId: getAgentJob
  path: GET /v1/agent/job/{id}
- operationId: listAgentJobs
  path: GET /v1/agent/jobs
- operationId: checkAuth
  path: GET /v1/auth
- operationId: createCrawlJob
  path: POST /v1/crawl
- operationId: getJob
  path: GET /v1/job/{id}
- operationId: cancelJob
  path: PUT /v1/job/{id}/cancel
- operationId: getJobMarkdown
  path: GET /v1/job/{id}/markdown
- operationId: getJobUrls
  path: GET /v1/job/{id}/urls
- operationId: resendJobWebhook
  path: POST /v1/job/{id}/webhook/resend
- operationId: createFeed
  path: POST /v2/feed
- operationId: getFeed
  path: GET /v2/feed/{id}
- operationId: deleteFeed
  path: DELETE /v2/feed/{id}
- operationId: getFeedJson
  path: GET /v2/feed/{id}/json
- operationId: pauseFeed
  path: PUT /v2/feed/{id}/pause
- operationId: resumeFeed
  path: PUT /v2/feed/{id}/resume
- operationId: getFeedRss
  path: GET /v2/feed/{id}/rss
- operationId: runFeed
  path: PUT /v2/feed/{id}/run
- operationId: resendFeedWebhook
  path: POST /v2/feed/{id}/webhook/resend
- operationId: listFeeds
  path: GET /v2/feeds
- operationId: getOrganizationUsage
  path: GET /v2/organization/usage
- operationId: getScrapeV2
  path: GET /v2/scrape/{id}
coverage:
  mcp_tools: 1
  rest_operations: 24
  rest_operations_with_tool: 1
  rest_only: 23
  mcp_only: 0
  percent_rest_covered: 4.2
gap: An agent on the MCP surface can scrape one page at a time but cannot start a crawl job, read job results, create
  or read a feed, run the crawling agent, or check balance/usage. The provider's own Claude Code skill (skills/webcrawlerapi-com-webcrawlerapi.md)
  and the webcr CLI (cli/) are how the crawl surface reaches agents today.