context_add_relation
ChatGPTRelate a task to another document in the same nest. type is blocks | blocked_by | relates_to | duplicates | duplicated_by, read from the task's side ("this task blocks <node>"). blocked_by is the inverse view of blocks, not a separate link. Refuses self-relations, duplicates, and a blocks link that would make a loop. Needs edit access to the task. On this server-level endpoint, nest (id, name or slug) is required.
context_board_items
ChatGPTList a saved board's items — the same rows the board shows (its tasks-only, tag and completed filters), grouped into the board's columns in board order. Each row has title, column(s), status, assignee, priority, due, parent, tags and timestamps. Sortable and paged. On this server-level endpoint, nest (id, name or slug) is required.
context_boards
ChatGPTList this nest's saved boards (shared ones plus your private ones): id, name, what the columns are (folder, status, tag, assignee or priority), swimlanes and the board's filters. Use an id with context_board_items or context_bulk_update. On this server-level endpoint, nest (id, name or slug) is required.
context_bulk_update
ChatGPTApply one change to up to 100 nodes: move them into a board column (a done column completes tasks), add or remove tags, set priority, assignee or due date. Each node is its own governed edit with its own version. A node whose change would need review (prime, or governed without a bypass) or that you can't edit is skipped and reported, never saved as a draft. Returns a result per node. On this server-level endpoint, nest (id, name or slug) is required.
context_create
ChatGPTCreate a new knowledge node in the vault (type "task" for a board task; any node may carry assignee/due/priority/parent). The title must not already be in use — these tools resolve documents by title, so a duplicate is refused rather than filed as an unreachable twin. To change an existing document use context_update; to refile it use context_move. On this server-level endpoint, nest (id, name or slug) is required.
context_delete
ChatGPTDelete a node from the vault by title or id. Removes everything keyed to it: version history, review state, annotations, comments, watchers, schedules, trigger hooks, workflow edges, document grants, and any steward assignments scoped to it. Irreversible — the node must be recreated to come back, and it comes back clean: no history, no stewards, no shares. A node under open review is frozen unless you can approve it (owner, admin, or its reviewer). On this server-level endpoint, nest (id, name or slug) is required.
context_get
ChatGPTRead the full content of one document, by title or ID. On this server-level endpoint, nest (id, name or slug) is required.
context_list
ChatGPTBrowse vault contents with optional type, tag, status, assignee, priority, due_before, or limit filters. Tasks report assignee / due / priority / parent. On this server-level endpoint, omit nest to run it across ALL accessible nests (each hit names its nest).
context_move
ChatGPTMove a node to a different folder — or, with from_folder, rename or move a whole folder (owner/admin only; sub-folders and every document under it follow, ids change, history and review state survive). Moving a node: the id changes; version history, review state, annotations, comments, watchers, schedules, links, and document grants all follow. The title and its [[wiki-link]] stay valid. A node under open review is frozen unless you can approve it (owner, admin, or its reviewer). On this server-level endpoint, nest (id, name or slug) is required.
context_overview
ChatGPTGet a complete map of the vault: total node count, types, tags, and a title+snippet for every node. On this server-level endpoint, omit nest to run it across ALL accessible nests (each hit names its nest).
context_query
ChatGPTFind documents with a selector, a small query language over tags, types, titles and scopes. Supports: #tag, type:X, [[Title]], scope:X. Combine with +AND, |OR, -NOT. On this server-level endpoint, omit nest to run it across ALL accessible nests (each hit names its nest).
context_relations
ChatGPTList a task's relations — what it blocks, what blocks it (blocked_by), related tasks (relates_to), and duplicates — by title or ID. Says whether the task is blocked (an open blocker is one not in a done column). Related documents you can't read are counted as hidden, never named. On this server-level endpoint, nest (id, name or slug) is required.
context_remove_relation
ChatGPTRemove one of a task's relations by its relation id (from context_relations or context_get). Needs edit access to the task. On this server-level endpoint, nest (id, name or slug) is required.
context_request_access
ChatGPTAsk a nest owner/admin for write access when a write was refused — to one document, a folder, or the whole nest. Nothing changes until someone approves it; you're told the verdict. Use this when context_update / context_create / context_delete / context_move return a permission error that mentions it. Asking twice for the same thing returns the pending request instead of a second one. On this server-level endpoint, nest (id, name or slug) is required.
context_resolve
ChatGPTFetch the full text of the documents a selector matches, plus documents linked to them up to hops away (hops 0 for matches only). Stops at the first document past the token budget. Use context_query when you don't need the text. On this server-level endpoint, nest (id, name or slug) is required.
context_search
ChatGPTSearch all documents by keyword, across content, titles, tags and metadata — the same index-backed match as REST GET /search. Returns count (everything found) and the first limit hits, title matches first. On this server-level endpoint, omit nest to run it across ALL accessible nests (each hit names its nest).
context_submit_review
ChatGPTSubmit a node for steward review. The node will be marked as pending review and stewards will be notified. On this server-level endpoint, nest (id, name or slug) is required.
context_task_log
ChatGPTRegister what you did on a task — one line in its activity timeline, attributed to you (and to the API key and agent you connected with). Use it to say you picked a task up (status "working"), finished it ("done") or are stuck ("blocked"). claim: true also makes you the assignee if nobody is. Needs edit access to the task; it does not change the task's content or version. On this server-level endpoint, nest (id, name or slug) is required.
context_update
ChatGPTUpdate an existing node — replace or append content, set tags, set task fields (assignee, due, priority, parent), move it through its lifecycle. Whether the edit publishes is decided by this nest's stewardship rules, not by the caller. On this server-level endpoint, nest (id, name or slug) is required.
context_upload
ChatGPTUpload a file into the nest by URL or as base64 — no multipart. Routed by the file's extension: images (png, jpg, jpeg, gif, webp; 10 MB) and video (mp4, webm; VIDEO_MAX_MB, default 30 MB) go to the nest's asset store and come back as ready-to-paste markdown (an image reference, or a video URL on its own line) to put in a document with context_create/context_update; a PDF (PDF_MAX_MB, default 25 MB) becomes a governed pdf document with its text extracted; a .pptx, .docx or .xlsx (OFFICE_MAX_MB, default 25 MB) becomes a governed markdown document (a section per slide with speaker notes, the Word document's headings, lists and tables, or a table per spreadsheet sheet) with the original kept as a download. Documents follow the nest's review rules like context_create (a draft in a governed nest). SVG and other types are refused. Send exactly one of content_base64 (with filename) or url (https only, fetched by the server; private and internal addresses are refused and redirects are not followed). base64 travels inside this MCP request, which the server caps (MAX_BODY_BYTES, default 10 MB, so about 7.5 MB of file) — use url for anything bigger. On this server-level endpoint, nest (id, name or slug) is required.
context_versions
ChatGPTShow the version history of a node — all edits, who made them, their approval status, and which version is currently approved for AI use. On this server-level endpoint, nest (id, name or slug) is required.
nest_create
ChatGPTCreate a new nest owned by the caller. Returns the new nest's id — target it with the other tools afterwards.
nest_index
ChatGPTList every nest these credentials can read, with name, id, permission and document count. Call this FIRST.