Next Focus AI

Showing 25 of 77 from the live probe.

• Alignment Signal Query

  Query alignment signals for a scope with filtering and analysis capabilities. Returns detected configuration alignment opportunities or concerns.

• Alignment Mode Status

  Check current operational mode and trust metrics for a scope. Returns mode (STANDARD, TRUSTFIRST_REVIEW, AUTO_PILOT), trust scores, and eligible transitions.

• Alignment Task Action

  Take action on an alignment task (accept, reject, defer, or dismiss). Requires the task ID and the action key from the task's action list.

• Alignment Institutional Query

  Query institutional intelligence and historical decisions for alignment context. Returns supporting and counter evidence from the Steward memory pipeline.

• Nf Webpilot Run Recipe

  Execute a bounded browser recipe (navigate/click/fill/extract/assert) against a governed, sandboxed browser session and return per-step results + extracted text. Runs in an ephemeral leased VM (no persistent state). Requires 'execute' permission in the given scope; navigated URLs must be public (private/loopback targets are rejected).

• Nf Journey List

  List journeys with filtering by type, status, and scope. Returns journey summaries including title, type, approval status, and execution stats.

• Nf Journey Execute

  Execute a journey programmatically in headless or agent-driven mode. Returns execution ID and initial status. Subscribe to journeyExecutionProgress for real-time updates.

• Nf Journey Status

  Check the status of a journey execution. Returns current execution status, step progress counters, timing, and error log.

• Nf Journey Create

  Create a journey by assembling existing segments. Returns the new journey ID.

• Nf Journey Suggest

  Get journey suggestions based on the user's current context. Uses semantic search and FTS across journey titles and descriptions.

• Get Scope

  Retrieve a single scope by ID with full metadata including hierarchy type, depth, ancestor chain, and governance domain.

• List Scopes

  List scopes accessible to the current user with optional filtering by scope_type, hierarchy_type, direct parent, and active status.

• Get Scope Chain

  Get the full hierarchy chain for a scope, ordered [self, parent, ..., root]. Useful for understanding the organizational position of a scope.

• Get Scope Members

  Get members of a scope with their assigned roles. Returns member_of edges with user/agent IDs and role assignments.

• List Boards

  List Boards the current principal can access (owned or shared via a board_access grant), most-recently-updated first. Optionally filter by scope.

• Get Board

  Get a Board's metadata and its list of pages. Returns NOT_FOUND if the principal has no access (existence is hidden).

• Describe Board

  Semantic dump of a Board from the knowledge graph: its pages and, per page, the shapes with their type, text content, and media flags — the cheap answer to 'what is on this board?'. Reads BKI (may lag live edits by up to ~10s). Optionally restrict to a single page.

• Search Boards

  Find Boards and content. Returns `boards` — accessible boards whose TITLE matches the query (use this to locate a board by name) — AND `matches` — shapes whose text content matches via full-text search (notes, labels). Searches one board when board_id is given, otherwise all accessible boards.

• Create Board

  Create a NEW empty Board in the agent's data scope, owned by the requesting user. The board's internal room/storage identifiers are generated server-side — NEVER pass room_id or storage_key. The requesting agent is automatically granted write access to the board it creates, so it can immediately populate it (e.g. with create_shapes) using the returned board_id. Attended chat turns only. Use this instead of create_item when the user asks to create a board. If the user's request names the board —

• Create Shapes

  Author shapes onto a Board as the agent (a conflict-safe live peer). Pass one text per shape; shapes are auto-placed in a grid. Choose `shape`: use 'note' (sticky) ONLY for short labels (a few words) — notes are fixed-size and get cramped; use 'rectangle' for longer text or Yoga grid/table layouts (holds more text, resizable via width/height, supports aspect ratios); or a geo primitive (diamond, triangle, ellipse, star, hexagon, …) for variety. If `shape` is omitted the kind is auto-picked by te

• Update Shapes

  Modify EXISTING shapes on a Board: recolor, change text, convert shape kind (e.g. note → diamond, rectangle → triangle), and/or resize geo cards. Select shapes by shape_ids (from describe_board's tldraw_shape_id) or by a text `match` substring (e.g. 'chicken'). Converting kind preserves the text + position. Acts on the board's first page. Requires a write board_access grant; governance-gated.

• Delete Shapes

  Remove shapes from a Board's first page. Select by shape_ids (from describe_board's tldraw_shape_id) or a text `match` substring (e.g. delete the card about 'chickens'). Requires a write board_access grant; governance-gated via the DELETE verb. Removes board CONTENT only — it can never delete the board itself.

• Arrange

  Tidy a Board's shapes into a clean grid (Yoga flex-wrap layout). Rewrites positions of the box-like shapes on the board's FIRST page (arrows/connectors are left alone) — governance-gated like any write. Requires an agent principal with a write board_access grant.

• Frame

  Group a Board's FIRST-page box-like shapes into a single labeled frame (a titled container). Idempotent by title. Requires an agent principal with a write board_access grant.

• Visualize

  Compose structured data onto a Board as a titled, laid-out view — the PREFERRED way to build a board. When a user asks for a board about a topic without specifying a layout, LEAN toward a polished, colored, presentation-grade view — it reads far better than a loose scatter of notes. Pick the kind by SHAPE OF THE DATA: • A COMPARISON (N things measured across the SAME criteria — languages × features, vendors × capabilities) → kind='matrix' WITH a `rail` (the criteria) and one column per thing: a