6sense
Ask an AI agent about 6sense account insights, buying signals and keyword trends.
Search O'Reilly books, videos, courses and events from an AI agent.
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
The O'Reilly MCP server connects the O'Reilly learning platform's catalog — books, videos, courses, audiobooks and live events — to Claude, ChatGPT and any MCP-compatible agent. The connector Anthropic lists exposes one tool, a keyword search returning metadata and a link rather than book text. Sign-in is OAuth, or a revocable personal access token.
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.
Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.
scopes_supported is an empty array in both the RFC 9728 resource descriptor and the authorization server metadata, and the OpenAPI security definition declares an empty scopes object. Nothing at the OAuth grant narrows anything, and a token cannot be scoped down. On a single read-only tool the exposure is small; the mitigation is that tokens are individually revocable.tools/list response was seen. Everything about the tool's parameters comes from O'Reilly's published specification, not from the server.languages is set explicitly.ask_oreilly_experts on the paid tier is the answer to that, and it is not in this listing./apidocs/ returns the same shell. A plain fetch of the documented URL yields four words of text; the content lives in OpenAPI documents under a separate /dist/ path that required a real browser to discover. Anything summarising these docs from a simple fetch has read nothing.content-discovery path and O'Reilly's documented search path are both live with separate resource descriptors, and neither party's documentation acknowledges the other's URL.Not the one Anthropic lists. Its response schema carries title, authors, publisher, page count, a cover image and a link — no field for body text, chapters or excerpts. A separate paid Expert server does return snippet text through a citation tool, but that is a different endpoint requiring a different entitlement.
None at all. Both the RFC 9728 resource descriptor and the authorization server metadata declared scopes_supported as an empty array on 2026-08-23, and the OpenAPI security definition declares an empty scopes object too. Three independent declarations agree. Nothing can be granted selectively, which matters less on a single read-only tool.
Yes. O'Reilly states that all requests require authentication and that no anonymous requests are permitted, and an anonymous call returned HTTP 401 when we checked on 2026-08-23. O'Reilly's guide describes the Search server as available to all O'Reilly members, so it is a membership benefit rather than a public API.
Its Membership Agreement forbids it in general terms. Content accessed via the Service may not be stored, reproduced or transmitted, including as input or a prompt to a machine learning or AI system, without written permission from the copyright owner. The MCP server is O'Reilly granting that permission through a controlled product rather than an exception you may assume.
No, and its robots.txt names the crawlers. GPTBot, anthropic-ai, Google-Extended, CCBot and cohere-ai sit under a heading reading AI training scrapers with content access disallowed. The Membership Agreement separately forbids extracting content to train a machine learning or AI system. O'Reilly treats attributed retrieval and model training as different things.
No. The single listed tool is search_oreilly_content, a keyword search whose documented parameters are all query filters, with no identifier of a record to modify. Nothing creates, updates or deletes. Even the paid Expert server adds only two more read tools, so no O'Reilly MCP surface writes anything at all.
No single call spends anything metered. O'Reilly publishes no per-call price, credit balance or quota, and the membership you already hold is what grants access. No tool reports usage or billing either, but with nothing consumable per call there is no runaway spend for an agent to cause.
Because two paths are live. Anthropic's directory registers the content-discovery path while every O'Reilly document gives a search path instead. We confirmed on 2026-08-23 that both return an identical authentication challenge and each publishes its own resource descriptor. Neither redirects to the other, and O'Reilly never mentions the directory's path.
robots.txt with its labelled bot groups (retrieved 2026-08-23) · retrieved 2026-08-23www-authenticate challenge, RFC 9728 metadata at https://api.oreilly.com/.well-known/oauth-protected-resource/api/content-discovery/v1/mcp, and authorization server metadata at https://api.oreilly.com/.well-known/oauth-authorization-server/api/content-discovery/v1 (2026-08-23) · retrieved 2026-08-23https://api.oreilly.com/api/search/v1/mcp with its own resource descriptor (2026-08-23) · retrieved 2026-08-23Connect O'Reilly once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.