Browse

Rerank a list of documents by relevance to a query using Jina AI's reranking API. Requires JINA_API_KEY.

Settle your review of the candidate output. Must be called exactly once, when your review is complete.

Read a project card from the knowledge vault: type, status, next step, updated date, and the places on disk it covers (repo, folder, workflow). Cards are the only home of status. Without a name, returns the working directory's project card (created on first use).

Set a card's status and/or next step and stamp its updated date. This is the only place status and the next step are written: never put them in a note. Without a name, updates the working directory's project card (created on first use). Status archived moves the project folder to archive/. add_search_folder links a folder outside the project (docs, a shared library) so recall and search_local cover it too, and indexes it.

Save a new note in the knowledge vault: one idea per note, with frontmatter (type, status, created, summary, tags, links). Refuses a title that already has a note - one home per fact, so edit that note instead. Use for a decision, a gotcha, a concept or a tool worth remembering beyond this session; it goes in the project's folder unless scope says otherwise. Status and next steps go to the card instead.

Find the notes in the knowledge vault related to a query, by walking the vault's link graph from the words of the query (no model involved). Returns note paths with their one-line summaries, best first, within a small token budget. Each turn already carries a recall for the user's request; call this to look up something else (a project, a decision, an error). Then read only the notes you need with read_file.

List files and directories in a given directory path. Returns names and types (file/dir). Use to discover project structure before reading or editing files. Omit path to list the current working directory.

Load a document file and return its content as text. Supports PDF (langchaingo pdf loader), DOCX/EPUB/RTF/ODT (pandoc), HTML (goquery, lynx), CSV, text, directory, notion, markdown (textutil). Use for reading documents into the conversation for analysis or RAG pipelines. Returns document content with optional splitting into chunks. Requires: source (absolute file path). Optional: type (pdf, csv, html, text - auto-detected from extension), columns (CSV column filter), password (encrypted PDF), chunkSize (split into chunks of N characters), chunkOverlap, chunkSplitter (recursive, token, markdown).

Load the full instructions for a named skill. Call this when the user's task matches one of the skills listed in <available_skills>. Returns the skill's complete instructions to follow.