> ## Documentation Index > Fetch the complete documentation index at: https://docs.gtm-api.com/llms.txt > Use this file to discover all available pages before exploring further. # Scrape LinkedIn Learning course search > One page of a LinkedIn LEARNING course search, the sixth row kind here: `filters` OR a pasted /search/results/learning/ `url` (that screen, NOT /courses/), exactly one of the two. filters is a REQUIRED keywords plus two closed vocabularies, difficulty and time_to_complete, whose snake_case values the NODE maps onto LinkedIn's own labels, so never send a label yourself; the screen's software and subject facets need the url form. Rows are courses keyed by course_slug (this wire has no numeric id), with a tracking-free /learning/ URL, title, instructor, a normalised duration, the release and viewer lines as printed, and artwork. WARNING, the English dependency is TOTAL, as on groups: the title is read out of the "Save the course X" button label and a card without a readable title is DROPPED, so a non-English executor returns ZERO rows at success. Page on has_more, never on paging.total. Contract: - MCP tool `scrape_linkedin_search_courses`, registry package `mcp.linkedin/linkedin_scraping`, mount `linkedin.scraping`. - Operation `action`, response envelope `action`. - Flags: none. ## OpenAPI ````yaml /api-reference/linkedin/openapi.yaml post /api/linkedin-scraping/search-courses openapi: 3.0.3 info: title: 'GTM API public contract: gtm.service.linkedin' description: >- Connected LinkedIn accounts and everything driven through them: account health and smart limits, conversations and messages, the connection graph, outbound posting, scraping, profile and company enrichment, and the antidetect browsers that execute it all. GENERATED. This document is projected from the Zod MCP tool registry in `product/mcp/gtm.mcp` (one tool per public endpoint, 1:1). Do not edit it by hand; edit the tool definition and regenerate with `pnpm openapi:public`. Surface: the public `/api` contract of `gtm.service.linkedin`, 159 operations. This is the only OpenAPI document the platform publishes. Internal (`/internal`) and health endpoints are deliberately absent: they are not part of any contract, they can change without notice, and the service source is their only description. Conventions: - Auth is a bearer JWT, optionally narrowed by the `Team-SID` header. - Every success body is an MCP envelope: `success: true` plus one typed `operation` shape (`search`, `get`, `create`, `update`, `delete`, `metrics`, `group_by`, `action`), and a `meta` block with `trace_id` for support. - Every failure is the same `McpError` envelope with a code from a fixed 16-code taxonomy, so a client maps errors once. - Lists page with `page_size` (0 to 500, default 50) plus an opaque forward `cursor`; `page_size: 0` returns counts only. - On `GET` and `DELETE`, object-valued query parameters (`filter`, `sort`) travel as JSON text and array-valued ones repeat as `name[]=value`. - The MCP-only `_meta` field (usage analytics) never reaches the backend and is not part of this contract. version: '1.0' contact: name: GTM API url: https://gtm-api.com email: support@gtm-api.com license: name: Proprietary url: https://gtm-api.com/license servers: - url: https://app.gtm-api.com/linkedin/v4 description: Production, through the app.gtm-api.com gateway security: - BearerJwt: [] TeamSid: [] tags: - name: antidetect_browser_logs description: >- Registry package `mcp.linkedin/antidetect_browser_logs`, served on MCP mount `linkedin.browsers`. - name: antidetect_browser_proxies description: >- Registry package `mcp.linkedin/antidetect_browser_proxies`, served on MCP mount `linkedin.browsers`. - name: antidetect_browsers description: >- Registry package `mcp.linkedin/antidetect_browsers`, served on MCP mount `linkedin.browsers`. - name: cloud_browser_sessions description: >- Registry package `mcp.linkedin/cloud_browser_sessions`, served on MCP mount `linkedin.browsers`. - name: cloud_browsers description: >- Registry package `mcp.linkedin/cloud_browsers`, served on MCP mount `linkedin.browsers`. - name: data_requests description: >- Registry package `mcp.linkedin/data_requests`, served on MCP mount `linkedin.data`. - name: linkedin_account_activity_log description: >- Registry package `mcp.linkedin/linkedin_account_activity_log`, served on MCP mount `linkedin.account-monitor`. - name: linkedin_account_block_log description: >- Registry package `mcp.linkedin/linkedin_account_block_log`, served on MCP mount `linkedin.account-monitor`. - name: linkedin_account_quota_hits description: >- Registry package `mcp.linkedin/linkedin_account_quota_hits`, served on MCP mount `linkedin.account-monitor`. - name: linkedin_account_smart_limits description: >- Registry package `mcp.linkedin/linkedin_account_smart_limits`, served on MCP mount `linkedin.accounts`. - name: linkedin_account_snapshots description: >- Registry package `mcp.linkedin/linkedin_account_snapshots`, served on MCP mount `linkedin.account-monitor`. - name: linkedin_account_sync_runs description: >- Registry package `mcp.linkedin/linkedin_account_sync_runs`, served on MCP mount `linkedin.account-monitor`. - name: linkedin_accounts description: >- Registry package `mcp.linkedin/linkedin_accounts`, served on MCP mount `linkedin.accounts`. - name: linkedin_auto_scrape_results description: >- Registry package `mcp.linkedin/linkedin_auto_scrape_results`, served on MCP mount `linkedin.auto-scrapes`. - name: linkedin_auto_scrape_runs description: >- Registry package `mcp.linkedin/linkedin_auto_scrape_runs`, served on MCP mount `linkedin.auto-scrapes`. - name: linkedin_auto_scrapes description: >- Registry package `mcp.linkedin/linkedin_auto_scrapes`, served on MCP mount `linkedin.auto-scrapes`. - name: linkedin_benchmarks description: >- Registry package `mcp.linkedin/linkedin_benchmarks`, served on MCP mount `linkedin.account-monitor`. - name: linkedin_connection_invitations description: >- Registry package `mcp.linkedin/linkedin_connection_invitations`, served on MCP mount `linkedin.network`. - name: linkedin_connection_requests description: >- Registry package `mcp.linkedin/linkedin_connection_requests`, served on MCP mount `linkedin.network`. - name: linkedin_connections description: >- Registry package `mcp.linkedin/linkedin_connections`, served on MCP mount `linkedin.network`. - name: linkedin_conversations description: >- Registry package `mcp.linkedin/linkedin_conversations`, served on MCP mount `linkedin.messaging`. - name: linkedin_custom_requests description: >- Registry package `mcp.linkedin/linkedin_custom_requests`, served on MCP mount `linkedin.platform`. - name: linkedin_enrichment description: >- Registry package `mcp.linkedin/linkedin_enrichment`, served on MCP mount `linkedin.enrichment`. - name: linkedin_followers description: >- Registry package `mcp.linkedin/linkedin_followers`, served on MCP mount `linkedin.network`. - name: linkedin_messages description: >- Registry package `mcp.linkedin/linkedin_messages`, served on MCP mount `linkedin.messaging`. - name: linkedin_posting description: >- Registry package `mcp.linkedin/linkedin_posting`, served on MCP mount `linkedin.content`. - name: linkedin_scraping description: >- Registry package `mcp.linkedin/linkedin_scraping`, served on MCP mount `linkedin.scraping`. paths: /api/linkedin-scraping/search-courses: post: tags: - linkedin_scraping summary: Scrape LinkedIn Learning course search description: >- One page of a LinkedIn LEARNING course search, the sixth row kind here: `filters` OR a pasted /search/results/learning/ `url` (that screen, NOT /courses/), exactly one of the two. filters is a REQUIRED keywords plus two closed vocabularies, difficulty and time_to_complete, whose snake_case values the NODE maps onto LinkedIn's own labels, so never send a label yourself; the screen's software and subject facets need the url form. Rows are courses keyed by course_slug (this wire has no numeric id), with a tracking-free /learning/ URL, title, instructor, a normalised duration, the release and viewer lines as printed, and artwork. WARNING, the English dependency is TOTAL, as on groups: the title is read out of the "Save the course X" button label and a card without a readable title is DROPPED, so a non-English executor returns ZERO rows at success. Page on has_more, never on paging.total. Contract: - MCP tool `scrape_linkedin_search_courses`, registry package `mcp.linkedin/linkedin_scraping`, mount `linkedin.scraping`. - Operation `action`, response envelope `action`. - Flags: none. operationId: scrape_linkedin_search_courses requestBody: required: false content: application/json: schema: $ref: '#/components/schemas/ScrapeLinkedinSearchCoursesRequest' responses: '200': description: '`action` success envelope.' content: application/json: schema: $ref: '#/components/schemas/ScrapeLinkedinSearchCoursesResponse' 4XX: $ref: '#/components/responses/McpClientError' 5XX: $ref: '#/components/responses/McpServerError' components: schemas: ScrapeLinkedinSearchCoursesRequest: type: object description: Request body of `scrape_linkedin_search_courses`. properties: linkedin_account_sid: type: string nullable: true minLength: 18 maxLength: 18 pattern: ^ln_ac_ description: >- Executor account (ln_ac_...). OPTIONAL everywhere on this surface. Given: the call runs on that account ONLY; a saturated or held scraping bucket refuses 429 bucket_saturated (with retry_after). Omitted: the service auto-picks one of your connected accounts with remaining capacity (SN verbs pick only Sales-Navigator seats; 422 no_connected_accounts when none is ready, 429 when all are at capacity). idempotency_key: type: string nullable: true maxLength: 128 description: >- Ledger replay guard: a repeat call with the same (team, key) returns the stored outcome; no re-execution. Recommended on every search run. The KEY ALONE decides: the probe does not compare arguments, so reusing one key after changing the arguments hands back the FIRST result. On the twelve search verbs that matters twice over, because switching a call from filters to url (or back) under one key replays instead of running the new search. New search, new key. url: type: string nullable: true maxLength: 2048 pattern: ^https:\/\/www\.linkedin\.com\/search\/results\/learning\/ description: >- EXCLUSIVE with `filters`: send url OR filters, never both (422) and never neither (422). A search URL built in the LinkedIn UI, and the escape hatch for everything the filter vocabulary cannot express. MUST start with https://www.linkedin.com/search/results/learning/ : a URL from another LinkedIn search screen is refused here and again by the backend (422 invalid_search_url), because running it would silently scrape the wrong thing. On the url half OUR number wins: the offset or page parameter baked into the pasted URL is overwritten, while every other parameter rides through untouched. filters: type: object nullable: true properties: keywords: type: string maxLength: 256 description: >- REQUIRED: this engine has no wire form without a query, so a filter object carrying only a difficulty or only a length band is refused rather than run wide. difficulty: type: array nullable: true items: type: string enum: - beginner - intermediate - advanced maxItems: 3 description: >- Course level, LinkedIn Learning's own three. Send these readable names: the node maps each onto LinkedIn's own label ("Beginner") while it composes the URL, so a label sent from here is refused by its validator. time_to_complete: type: array nullable: true items: type: string enum: - under_10_mins - 10_to_30_mins - 30_to_60_mins - 1_to_2_hours - 2_to_3_hours - 3_plus_hours maxItems: 6 description: >- Course-length bands, in LinkedIn's own dropdown order. Readable names only, mapped node-side exactly like difficulty ("< 10 mins", "3+ hours"); the bands are what the UI offers, so an arbitrary minute range cannot be asked for. required: - keywords additionalProperties: false description: >- EXCLUSIVE with `url`: send filters OR url, never both (422) and never neither (422). LinkedIn LEARNING course-search filters. keywords is REQUIRED; difficulty and time_to_complete are closed vocabularies whose values the NODE translates into LinkedIn's own labels, so send the names spelled here and never a LinkedIn label. Any other key is REFUSED rather than ignored, which matters because the learning screen also ships software and subject facets this vocabulary does not carry: those are reachable only by building the search in the LinkedIn UI and pasting the URL into the url field. page: type: integer minimum: 1 maximum: 100 description: >- LinkedIn page number, default 1; ONE page per call, so re-call with page + 1 while paging.has_more. On the url half OUR number wins: the offset or page parameter baked into the pasted URL is overwritten, while every other parameter rides through untouched. ScrapeLinkedinSearchCoursesResponse: type: object properties: success: type: boolean enum: - true operation: type: string enum: - action action: type: string description: kebab-case verb; matches the route segment. item: {} result: type: object properties: rows: type: array items: type: object properties: course_slug: type: string description: >- The /learning/{slug} path segment, and the ONLY identity a course row carries: this vertical has no numeric id anywhere on the wire. A card whose slug cannot be read is dropped rather than returned half-formed, and a repeat of a slug already seen on this page is dropped too. course_url: type: string description: >- https://www.linkedin.com/learning/{course_slug}/ , composed from the slug, so it is stable and free of tracking parameters. title: type: string description: >- Course title, read out of the English "Save the course " button label. A card whose title cannot be read is dropped, so this is never null, which is exactly why a non-English executor returns an empty page instead of untitled rows. author: type: string nullable: true description: >- The instructor, from the card's "By: <name>" line with the prefix stripped. A bare NAME: no member id, no vanity slug, so joining an instructor back to a LinkedIn profile needs a second search. duration_text: type: string nullable: true description: >- Course length, NORMALISED rather than quoted: the parser reads LinkedIn's accessible form ("6 hours and 14 minutes", "1 day and 10 hours"), folds days into hours and re-prints it compactly ("34h 10m", "45m"). It is not the string on the card and not a number of minutes, so parse it if you need arithmetic. released_text: type: string nullable: true description: >- The release line VERBATIM, prefix included ("Released Jan 2026"). Free text, not a date, and unlike author the "Released " prefix is deliberately not stripped. viewers_text: type: string nullable: true description: >- The viewer counter exactly as printed ("12,345 viewers", "1.2M viewers"). Kept as text on purpose: LinkedIn rounds the compact form, so there is no honest integer to parse it into. thumbnail_url: type: string nullable: true description: >- Course artwork, widest rendition available. Best effort: a card renders several images and the most repeated asset is taken as the thumbnail. position: type: integer nullable: true description: >- The card's slot on THIS page, 0 based and page local, so it is not a rank across pages. Numbering happens before duplicate slugs are dropped, which can leave gaps. The parser always fills it, so null means the key was missing from the wire row. required: - course_slug - course_url - title - author - duration_text - released_text - viewers_text - thumbnail_url - position description: >- The transient result list, in LinkedIn’s order; NOTHING persisted. paging: type: object properties: page: type: integer page_size: type: integer has_more: type: boolean total: type: integer nullable: true required: - page - page_size - has_more - total description: >- Page-numbered paging: re-call with page + 1 while has_more. total carries LinkedIn's own result count when the page printed one, best effort and null otherwise, so drive the loop from has_more. data_request: type: object properties: {} description: >- The kind="scrape" DataRequest journal row for this call (terminal completed), embedded as result.data_request; served_from_cache always false. Full DataRequestDomain shape owned by ./data_requests.md, so it is left passthrough here. required: - rows - paging - data_request meta: type: object properties: trace_id: type: string description: UUID v7; same 128-bit value as the X-Trace-Id header. span_id: type: string pattern: ^[0-9a-f]{16}$ description: 16 hex chars, root span of this request. timestamp: type: string description: ISO 8601 UTC (Y-m-dTH:i:sZ), response time. duration_ms: type: integer minimum: 0 description: Server-side wall clock. debug_url: type: string description: Deep link to the post-call analysis UI. required: - trace_id - span_id - timestamp - duration_ms - debug_url required: - success - operation - action - item - result - meta McpError: type: object properties: success: type: boolean enum: - false error: type: object properties: code: type: string enum: - validation_failed - nothing_to_update - not_found - relation_not_found - invalid_transition - limit_exceeded - payment_required - duplicate_rejected - conflict - delete_blocked - unauthorized - forbidden - rate_limited - internal_error - service_unavailable - not_implemented message: type: string recoverable: type: boolean suggestion: type: string field_errors: type: object additionalProperties: type: array items: anyOf: - type: string - type: object properties: rule: type: string message: type: string required: - rule - message blockers: type: array items: type: object properties: type: type: string description: >- Machine-readable blocker type (active_flow, pending_tasks, …). severity: type: string enum: - hard - soft description: >- hard = external action required; soft = acknowledge is enough. description: type: string entity_sid: type: string nullable: true count: type: integer resolution: type: string description: 'Hard: tool name to call. Soft: code for acknowledge[].' resolution_hint: type: string required: - type - severity - description - entity_sid - resolution - resolution_hint context: type: object additionalProperties: {} required: - code - message - recoverable meta: type: object properties: trace_id: type: string description: UUID v7; same 128-bit value as the X-Trace-Id header. span_id: type: string pattern: ^[0-9a-f]{16}$ description: 16 hex chars, root span of this request. timestamp: type: string description: ISO 8601 UTC (Y-m-dTH:i:sZ), response time. duration_ms: type: integer minimum: 0 description: Server-side wall clock. debug_url: type: string description: Deep link to the post-call analysis UI. required: - trace_id - span_id - timestamp - duration_ms - debug_url required: - success - error responses: McpClientError: description: >- MCP error envelope. `error.code` is one of validation_failed, nothing_to_update, not_found, relation_not_found, invalid_transition, limit_exceeded, payment_required, duplicate_rejected, conflict, delete_blocked, unauthorized, forbidden, rate_limited, not_implemented. content: application/json: schema: $ref: '#/components/schemas/McpError' McpServerError: description: >- MCP error envelope with `error.code` internal_error or service_unavailable. content: application/json: schema: $ref: '#/components/schemas/McpError' securitySchemes: BearerJwt: type: http scheme: bearer bearerFormat: JWT description: >- Access token issued by gtm.service.id. Its `access_identity` claim carries `team_sid`, `actor_sid` and `actor_type`, and that team scope is authoritative. TeamSid: type: apiKey in: header name: Team-SID description: >- Team scope for tokens that do not carry one. Ignored when the token already names a team. ````