Provider API Escape Hatch
Use provider-specific actions for common jobs, but treat them as shortcuts, not limits. When a first-class action cannot express the endpoint, filters, request body, pagination mode, or API version needed, use:
provider-api-catalogto inspect provider base URL, auth style, docs/spec URLs, placeholders, examples, and reusablecorpusRecipes.provider-api-docsto fetch registered provider docs/spec URLs.provider-api-requestto make the exact HTTP call. Credentials are injected server-side, private/internal URLs are blocked, and secrets are redacted.
Clay
Clay is a credentialed GTM provider capability, not a messaging channel. Use
provider clay; the server injects CLAY_PUBLIC_API_KEY with the official
clay-api-key header and restricts requests to the exact
https://api.clay.com origin, with /public/v0 as the default base path. Use
provider-api-docs for the registered official docs and OpenAPI spec on
developers.clay.com.
Clay searches are stateful company/people iterators, routines are asynchronous, and table queries are read-only, Enterprise-only, and require a known table id. The optional local Clay CLI/MCP browser login is separate from hosted Agent Native credentials. Do not install or vendor that plugin by default; its public repository currently declares no license.
Examples:
pnpm action provider-api-catalog --provider=hubspot
pnpm action provider-api-request --provider=hubspot --method=POST --path=/crm/v3/objects/deals/search --body='{"filterGroups":[{"filters":[{"propertyName":"products","operator":"CONTAINS_TOKEN","value":"Publish"}]}],"properties":["dealname","products","dealstage","closedate"],"limit":100}'
pnpm action provider-api-request --provider=bigquery --method=GET --path=/projects/{projectId}/datasets
Always cite provider, method, path, status, filters, row/sample count, and pagination/coverage gaps in analytical answers.
For source-record body searches across transcripts, messages, tickets, issues, notes, documents, or conversation logs, use the raw body endpoint or native provider search endpoint for that record type. Parent/container metadata such as call lists, titles, summaries, or briefs is discovery evidence only; it cannot support a "no mentions" or complete body-text claim.
Example corpus recipe:
pnpm action provider-corpus-job --operation=start --mode=batch-search --request='{"provider":"gong","method":"POST","path":"/calls/transcript","body":{"filter":{"callIds":[]}}}' --batch='{"inputDatasetId":"<staged-call-id-dataset>","inputValuePath":"id","batchSize":20,"itemBodyPath":"filter.callIds","responseItemsPath":"callTranscripts"}' --search='{"queries":["Figma MCP","model context protocol"],"textPaths":["transcript"],"idPaths":["callId"]}'