Coursera

Search your organisation's Coursera catalogue and run an AI Role Play scenario in chat.

Add to Agentman Studio

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

Resources

What the Coursera connector does

The Coursera MCP server brings a workplace learning catalogue into Claude, ChatGPT or any MCP-compatible agent. Four tools search courses, videos, hands-on projects and course material; a fifth runs an AI Role Play practice scenario inside the conversation. Sign-in is OAuth. Coursera frames the connector around enterprise learners and their organisation's catalogue.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

Coursera tools (5)

  • coursera_roleplay_practice
  • search_courses
  • search_videos
  • search_hands_on_learning
  • get_course_material

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Enterprise gating is undocumented but strongly indicated. Coursera frames the connector for enterprise learners and organisation catalogues; Anthropic marks the entry commercial. No personal-account path is documented.
  • The vendor documentation URL is dead. business.coursera.help/hc/articles/000002420 returns 404, as does the whole admin- article namespace to an anonymous request.
  • resource_documentation is dead too. Both OAuth descriptors point at coursera.org/api/docs/mcp, which serves Coursera's API routing error page.
  • No vendor tool list exists. All five tool names come from Anthropic's directory snapshot of 2026-08-16 and are corroborated by no Coursera-published source.
  • No safety annotations were readable. The endpoint returns 401 to an anonymous handshake, so the read/write split on this page is inferred, not observed.
  • No per-tool scope map. One scope covers the whole surface; you cannot grant search without granting Role Play.
  • Prompts are unconfirmed. Anthropic's snapshot lists none, and the gated endpoint blocked our own check.
  • Role Play privacy is documented for the in-course feature, not for the connector. Coursera's statements are about Role Play activities; it has published nothing distinguishing a Role Play launched through MCP.
  • Role Play does not count for anything. Coursera states it is optional, ungraded, and not required for a course certificate — useful for practice, not for compliance evidence.
  • Language coverage is bounded. Coursera documents Role Play text chat in 22 languages and voice chat in 17 of them; voice chat is unsupported in the mobile app.
  • Content is licensed for personal, non-commercial use. Coursera's Terms grant a limited personal licence and prohibit reproducing or reselling Services content without written permission.

Frequently asked questions

Can an individual Coursera learner use the Coursera MCP server?

Probably not on a personal account. Coursera's launch announcement frames the connector around enterprise learners and their organisation's catalogue, and Anthropic's directory marks the entry commercial-visibility. Coursera has published no page describing personal-account setup, and the admin help article it links for access requirements sits behind a login wall. Treat it as a Coursera for Business feature until Coursera says otherwise.

Does the coursera_roleplay_practice tool write anything my employer can see?

Your conversation is not visible to your employer. Coursera's Role Play help article states that Role Play conversation history is private, not visible to your instructor, and not accessible to your organisation's admin if you are in an organisation's learning program. Coursera also states Role Play is optional and ungraded, so nothing is added to a certificate record.

Why does the Coursera connector need write permission when four of its tools are searches?

Because Role Play is an interactive session, not a query. Four tools read the catalogue; `coursera_roleplay_practice` runs a live conversation with an AI persona and returns feedback, which Coursera's directory entry flags as an interactive MCP app rather than a plain tool call. Anthropic labels the connector Read and write. Coursera publishes no per-tool permission map.

What does get_course_material actually return from a Coursera course?

Coursera does not document it, so we cannot say. Anthropic's connector page describes pulling key concepts from course modules into the conversation, which suggests extracted concepts rather than full lecture transcripts or video. Coursera's Terms of Use grant only a personal, non-commercial licence to its content and prohibit reproducing it. Assume extracts, not redistributable material.

Does Coursera document more MCP tools than the five Anthropic lists?

No. We found no Coursera-published tool list at all. Coursera's launch blog post describes the connector in prose and names no tools; the admin help article it points to returns 404 to the public. The five names come solely from Anthropic's directory snapshot of 2026-08-16, and the gated endpoint blocked our own tools/list check.

Why does the Coursera MCP server advertise two different OAuth scopes?

Because two OAuth stacks answer on one host. The RFC 9728 descriptor at the path-append form names `access_mcp_integration`, and the server's own challenge header asks for exactly that scope. The root form of the same descriptor names `access_chatgpt_widget_mcp` and a different authorisation endpoint. For MCP clients, the path-append form is the one that applies.

Sources

  • Coursera Blog, "Skill Building in the Flow of Work: Coursera Connector for Claude" — https://blog.coursera.org/coursera-connector-for-claude/ (published 2026-06-17, fetched 2026-08-19). This is the only public Coursera page describing the connector · retrieved 2026-06-17
  • Coursera Help Center, "Practice your skills with Role Play" — https://www.coursera.support/s/article/learner-000002340?language=en_US (article dated 2026-05-20, fetched 2026-08-19; the site renders via JavaScript and served text to a crawler user-agent) · retrieved 2026-05-20
  • Coursera Terms of Use — https://www.coursera.org/about/terms (effective 2026-01-01, fetched 2026-08-19) · retrieved 2026-01-01
  • Anthropic connector page for Coursera — https://claude.com/connectors/coursera (fetched 2026-08-19) · retrieved 2026-08-19
  • Dead vendor documentation — https://business.coursera.help/hc/articles/000002420, the URL Anthropic's directory publishes, redirects to https://www.coursera.support/s/article/admin-000002420 and returns HTTP 404 anonymously (checked 2026-08-19). Four neighbouring admin- article IDs returned 404 identically, while learner- articles on the same host returned text · retrieved 2026-08-19
  • www.coursera.support/robots.txt is the Salesforce Experience Cloud default, Allow: /, with no Content-Signal line. www.coursera.org/robots.txt allows /llms.txt and disallows /api/, /search and /organizations/; blog.coursera.org/robots.txt carries the same pattern and no Content-Signal (all fetched 2026-08-19) · retrieved 2026-08-19
  • Coursera machine index — https://www.coursera.org/llms.txt (fetched 2026-08-19; last reviewed 2026-08-13 per the file, and it names no MCP or connector page) · retrieved 2026-08-19
  • Live RFC 9728 protected-resource descriptors — https://api.coursera.com/.well-known/oauth-protected-resource/mcp and https://api.coursera.com/.well-known/oauth-protected-resource (both 200 with different scopes, fetched 2026-08-19; the path-insert form https://api.coursera.com/mcp/.well-known/oauth-protected-resource returns 200 with an empty body) · retrieved 2026-08-19
  • Live RFC 8414 authorisation-server metadata — https://api.coursera.com/.well-known/oauth-authorization-server and https://api.coursera.com/.well-known/oauth-authorization-server/mcp (both 200 with different scopes and endpoints, fetched 2026-08-19) · retrieved 2026-08-19
  • Live anonymous probe — 2026-08-19: POST to the endpoint returns HTTP 401 with www-authenticate: Bearer error="invalid_token", resource_metadata="…", scope="access_mcp_integration" · retrieved 2026-08-19
  • Anthropic connector directory — https://claude.ai/directory/03c911b8-9789-4cb1-949b-236f9d06b2da (snapshot 2026-08-16) · retrieved 2026-08-16

More Productivity connectors

  • 6sense

    Ask an AI agent about 6sense account insights, buying signals and keyword trends.

  • Adobe Workfront

    Find, create, update and delete Workfront work items from an AI agent.

Browse all MCP connectorsClaude and ChatGPT setup guideBrowse AI skills

Product names and logos belong to their owners. Agentman is not affiliated with a connector's maker unless the page says the connector is built and run by Agentman.

Use it in an agent

Put Coursera to work.

Connect Coursera once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Open in Agentman Studio
Call (650) 285-1019Our AI receptionist answers.