Get API capabilities

How agents should use this endpoint

• This endpoint is public and has no authentication dependency. It returns the registry-backed documentation projection, never caller-specific readiness/state; large and dynamic catalogs are projected through focused list/detail routes.
• view=full is the default; view=compact returns compact routing indexes. The endpoint supports conditional requests with ETag/If-None-Match (304 when unchanged) and gzip when accepted by the client. Responses include cache headers.
• Use column_types to detect runnable vs non-runnable and billable vs non-billable operations.
• Use filtering for run/estimate payload semantics and guardrails.
• Use automation.auto_run for the auto-run field name and allowed modes, including the column-owned config.runCondition contract.
• Use automation.providers for provider-specific setup contracts, including explicit structured-field path mapping for add_to_crm.
• Use automation.workflow_blueprints for machine-readable multi-step flows such as lead_finder -> add_to_sequence and llm items[] -> sync_to_table -> email_finder -> add_to_crm -> add_to_sequence.
• Use list_builds plus /api/list-builds for durable LinkedIn-sourced company and lead list builds. This is the default path for hundreds or thousands of companies or leads.
• Use search plus the /api/search/* endpoints for smaller provider-hidden public-web lookups. Use LinkedIn endpoints only when LinkedIn itself is the source of truth.
• Use linkedin.filter_catalog or GET /api/capabilities/linkedin/filters to discover LinkedIn/Sales Navigator filter fields before constructing extra_params.
• Use endpoints.list_columns, create_column, create_projection_columns, delete_column, run_column, and estimate_column_run for column API discovery.
• Use endpoints.get_operation_run and endpoints.list_operation_runs to recover/poll Research Table run state. Bulk-job endpoints remain for native executor diagnostics and domains that still own bulk jobs.
• Use the capability and catalog discovery reference for compact/full projections, operation and endpoint indexes, details, pagination, schema lookups, ETags, and provider filter metadata.
• Use the operation-runs reference for authenticated list/get/cancel lifecycle contracts, caller visibility, statuses, filters, and cancellation conflicts.
• Use endpoints.query_leads, query_leads_count, query_leads_stats, query_lead_research, get_lead_filter_metadata, and get_available_lead_research_tables for the public leads API.
• Use endpoints.*sequence* and endpoints.*task* for workflow route discovery.
• Use endpoints.list_sender_accounts and get_sender_account only when advanced sequence routing needs explicit sender account IDs. Normal enrollments should use assignedToUserId and the assignee's defaults.
• Use execution_policies.workflows for actor model and workflow behavior constraints.
• Use execution_policies.workflows.tasks.ai_draft.context_contract to discover requester-context requirements and the draft preflight endpoint.
• Use execution_policies.llm.output_contract to determine output type behavior by mode (agent vs basic).
• Use webhooks.table_ingest for ingest auth contract and payload shapes.
• Use imports.csv for safe CSV import capabilities (validate-only, verify-by-task, rollback-by-task, assertion keys).

Note: the response example below is an abridged full-view snapshot. api_version is generated from the current date, and operation/catalog contents can change with the backend registry; the live endpoint is the source of truth for those values.

Live readiness/state is separate: use authenticated endpoints such as GET /api/llm/company-context for caller-specific requester context and draft_readiness.

Security: this endpoint returns metadata only. It never returns webhook secrets/tokens.