{"authentication":{"description":"OAuth via dynamic client registration (RFC 7591) and protected-resource discovery (RFC 9728) against the mono authorization server (my.decisioninsights.ai); a static Bearer API key; or unauthenticated access at a lower, IP-throttled rate limit.","oauthProtectedResourceMetadataUrl":"https://mcp.decisioninsights.ai/.well-known/oauth-protected-resource","schemes":["oauth2","bearer","none"]},"capabilities":{"pushNotifications":false,"streaming":false},"defaultInputModes":["application/json"],"defaultOutputModes":["application/json"],"description":"Decision Insights' public research knowledge graph (companies, projects, published content, and glossary terms) exposed over the Model Context Protocol for AI agents and research tools.","documentationUrl":"https://app.decisioninsights.ai/mcp","mcp":{"endpoint":"https://mcp.decisioninsights.ai/api/mcp-public","tools":[{"name":"search_company","description":"Search the knowledge graph for companies by name, keyword, or domain. Optionally filter to companies we publish content about under a given topic/tag. Results are ranked by how much content covers each company (content_volume), so the most-covered companies surface first."},{"name":"get_company","description":"Get one company's full profile: description, location, size, the themes and topics its published content covers, the open-source projects it owns (ranked by how much content covers each; `projects_total` and `projects_truncated` say when the list is a subset), and its most recent content (each with a canonical link and a link back to the original source)."},{"name":"search_project","description":"Search the knowledge graph for projects/topics by name or keyword. Ranked by how much content is linked to each project (content_volume)."},{"name":"get_project","description":"Get one project/topic: its description, its owning company as a dereferenceable identifier (`owner_organization_id`, the @id that company's own DI.AI record publishes, alongside its name), the companies that appear in its content, and its most recent linked content."},{"name":"search_content","description":"Search the published articles by title, excerpt, or source. Results are ordered by publish date, most recent first. Optionally filter to the content Decision Insights classifies as best suited for a given job function; the classification is our judgment about someone else's writing, not the publisher's own statement of audience, and every classified result says so alongside the value. It is applied to 46% of published articles, so a filtered search returns only the classified subset."},{"name":"get_content","description":"Fetch one published article/summary: title, summary, its canonical link, the link back to the original source, provenance (source/brand/type), the subject company, the companies the article mentions (each with the @id its own DI.AI record publishes), tags, and projects. Mentioned companies come from automated name extraction over the article text and may include false positives — `mentions_extraction` on the response says so."},{"name":"search_glossary","description":"Search the glossary for topics/tags by name or keyword. A decisioninsights.ai term URL is resolved to the term it names rather than keyword-matched, and that result carries `matched_on: identity`. Everything else is ranked by match quality: an exact name or alias match first, then a name prefix, then a name substring, then terms that matched only in the definition; name A-Z within each tier. Every result carries `matched_on` naming which of those five it was. This is definition lookup, not a content-volume ranking."},{"name":"get_glossary","description":"Get one topic/tag (glossary entry): its definition, how much content carries it (content_volume), and the companies most associated with it."},{"name":"related_companies","description":"Find the peer companies for a given company. Two paths, never mixed in one response — read the `method` field, which names the one that produced it. CURATED (the normal case): when the company's decisioninsights.ai record publishes a peer set, this returns exactly that set, identical to the Related organizations in that page's JSON-LD, so the two public surfaces agree and can be joined on each peer's `id`. That set is computed from the organization's own curated topic profile over the listed directory and excludes its parent and subsidiaries; no similarity is computed and `similarity`/`shared_tags` are therefore empty. SIMILARITY (fallback, only when a company has no curated peers): ranks by IDF-weighted overlap between the topics of the ARTICLES this knowledge graph publishes about each company, which measures our editorial coverage rather than how similar the companies are — a heavily-covered company accumulates topics outside its own subject area and can rank on that breadth; those results carry the shared topics behind the match (most distinctive first). Neither path is a curated competitor or alternatives list. Industry is shown per result for context but does not filter or rerank. Each result carries the peer's `id` (the @id its own DI.AI record publishes). content_volume is shown for context but never affects the ranking."}],"transport":"streamable-http"},"name":"DI.AI Public Knowledge Graph","preferredTransport":"MCP","protocolVersion":"2025-06-18","provider":{"organization":"Decision Insights","url":"https://decisioninsights.ai"},"skills":[{"description":"Search the knowledge graph for companies by name, keyword, or domain. Optionally filter to companies we publish content about under a given topic/tag. Results are ranked by how much content covers each company (content_volume), so the most-covered companies surface first.","id":"search_company","name":"search_company","tags":["search","companies"]},{"description":"Get one company's full profile: description, location, size, the themes and topics its published content covers, the open-source projects it owns (ranked by how much content covers each; `projects_total` and `projects_truncated` say when the list is a subset), and its most recent content (each with a canonical link and a link back to the original source).","id":"get_company","name":"get_company","tags":["companies","profile"]},{"description":"Search the knowledge graph for projects/topics by name or keyword. Ranked by how much content is linked to each project (content_volume).","id":"search_project","name":"search_project","tags":["search","projects"]},{"description":"Get one project/topic: its description, its owning company as a dereferenceable identifier (`owner_organization_id`, the @id that company's own DI.AI record publishes, alongside its name), the companies that appear in its content, and its most recent linked content.","id":"get_project","name":"get_project","tags":["projects","profile"]},{"description":"Search the published articles by title, excerpt, or source. Results are ordered by publish date, most recent first. Optionally filter to the content Decision Insights classifies as best suited for a given job function; the classification is our judgment about someone else's writing, not the publisher's own statement of audience, and every classified result says so alongside the value. It is applied to 46% of published articles, so a filtered search returns only the classified subset.","id":"search_content","name":"search_content","tags":["search","content"]},{"description":"Fetch one published article/summary: title, summary, its canonical link, the link back to the original source, provenance (source/brand/type), the subject company, the companies the article mentions (each with the @id its own DI.AI record publishes), tags, and projects. Mentioned companies come from automated name extraction over the article text and may include false positives — `mentions_extraction` on the response says so.","id":"get_content","name":"get_content","tags":["content","articles"]},{"description":"Search the glossary for topics/tags by name or keyword. A decisioninsights.ai term URL is resolved to the term it names rather than keyword-matched, and that result carries `matched_on: identity`. Everything else is ranked by match quality: an exact name or alias match first, then a name prefix, then a name substring, then terms that matched only in the definition; name A-Z within each tier. Every result carries `matched_on` naming which of those five it was. This is definition lookup, not a content-volume ranking.","id":"search_glossary","name":"search_glossary","tags":["search","glossary"]},{"description":"Get one topic/tag (glossary entry): its definition, how much content carries it (content_volume), and the companies most associated with it.","id":"get_glossary","name":"get_glossary","tags":["glossary","definitions"]},{"description":"Find the peer companies for a given company. Two paths, never mixed in one response — read the `method` field, which names the one that produced it. CURATED (the normal case): when the company's decisioninsights.ai record publishes a peer set, this returns exactly that set, identical to the Related organizations in that page's JSON-LD, so the two public surfaces agree and can be joined on each peer's `id`. That set is computed from the organization's own curated topic profile over the listed directory and excludes its parent and subsidiaries; no similarity is computed and `similarity`/`shared_tags` are therefore empty. SIMILARITY (fallback, only when a company has no curated peers): ranks by IDF-weighted overlap between the topics of the ARTICLES this knowledge graph publishes about each company, which measures our editorial coverage rather than how similar the companies are — a heavily-covered company accumulates topics outside its own subject area and can rank on that breadth; those results carry the shared topics behind the match (most distinctive first). Neither path is a curated competitor or alternatives list. Industry is shown per result for context but does not filter or rerank. Each result carries the peer's `id` (the @id its own DI.AI record publishes). content_volume is shown for context but never affects the ranking.","id":"related_companies","name":"related_companies","tags":["companies","similarity"]}],"supportedInterfaces":[{"protocolBinding":"MCP","protocolVersion":"2025-06-18","transport":"MCP","url":"https://mcp.decisioninsights.ai/api/mcp-public"}],"url":"https://mcp.decisioninsights.ai/api/mcp-public","version":"1.1.0"}