Graphlit MCP Server

Slack messages, Google Drive files, and Jira issues can enter the same Graphlit project alongside local documents, web pages, and text. Ingestion converts documents and HTML pages to Markdown and transcribes audio and video. Connector tools offer recurring feeds, while completion checks track asynchronous ingestion. Website crawling adds pages to the knowledge base; web and podcast searches return results without ingesting them.

For finding material, retrieveSources returns ranked matches with filters for content type, ingestion recency, feeds, and collections. For ongoing questions, promptConversation searches across the knowledge base, preserves conversation history, and returns generated answers and any citations. Users can group content into collections and query previously stored conversations. Further tools extract JSON from text according to a supplied schema, find similar images, and publish text as audio or generate images from prompts, storing those outputs in the project. Notification tools can send text through Slack, email, webhooks, or Twitter/X.