Last updated September 29, 2026

MCP Tools

The MCP tools are the operations an agent can call through MCP access, each gated by the token's scope. Use this page to decide which scope an integration needs, and which calls you can take back.

Note

We recommend the architect plugin for anything that changes your workspace. Give any other client a Read token, or follow Let another client write.

How to read the tables

Column	Meaning
Scope	Read: every token gets it. Read + write: only a token with that scope
Discard undoes it	Whether discarding the token's working copy reverses what the tool changed. No means the change is permanent

delete_nodes reads Mostly: a discard restores the nodes and their edges. Child nodes stay moved out of a deleted node, and a layer board left without its node stays a standalone board.

Boards and layers

Tool	What it does	Scope	Discard undoes it
`get_board_tree`	Every layer board under a root, with its drill path	Read	—
`get_child_boards`	The layer boards a board's nodes drill into	Read	—
`get_parent_board`	The board whose node drills into this one	Read	—
`get_workboard_details`	A board with all its elements and styles	Read	—
`get_nodes`	A board's nodes, all or under one parent	Read	—
`get_edges`	A board's edges, filtered by node	Read	—
`find_elements`	Layers, nodes, edges and aspect rows under a root, by name	Read	—
`get_archetypes`	The organization's node and edge archetypes	Read	—
`get_system_view`	A view folded across the apps on a board, from each app's aspects	Read	—
`plan_grouping`	A review of a board's containment against how its elements connect	Read	—
`create_nodes`	Add nodes to a board	Read + write	Yes
`update_nodes`	Rename, describe, re-parent, retag or resize nodes	Read + write	Yes
`create_edges`	Connect nodes	Read + write	Yes
`update_edges`	Rename, describe or retag edges	Read + write	Yes
`apply_diagram_info`	Set a board's title, summary and notes	Read + write	Yes
`create_board`	Add a layer board under a node	Read + write	Yes
`delete_nodes`	Delete nodes and the edges wired to them	Read + write	Mostly (see above)
`delete_edges`	Delete edges	Read + write	Yes
`create_context_board`	Draw a standalone context board that answers a question	Read + write	No
`delete_context_board`	Delete a context board and everything on it	Read + write	No
`create_decision_board`	Add a decision board for an adopted decision record	Read + write	No
`propose_archetypes`	Propose new or changed archetypes for an admin to review	Read + write	No

Styling and layout

Tool	What it does	Scope	Discard undoes it
`style_signals`	The facts a styling pass plans from	Read	—
`match_icons`	Icons that match plain names	Read	—
`apply_semantic_styles`	Style nodes, edges and containers by role	Read + write	Yes
`apply_icon_shape_styles`	Set node icons and shapes	Read + write	Yes
`apply_composition`	Set a board's arrangement, reading direction and spacing	Read + write	Yes

Aspects

Tool	What it does	Scope	Discard undoes it
`list_aspects`	A board's aspect rows, across every kind or one	Read	—
`get_aspect`	One aspect row in detail	Read	—
`get_node_aspects`	The aspect rows a node owns or is referenced by	Read	—
`author_tables`	Add or change the intended database tables of an app board	Read + write	Yes
`author_endpoints`	Add or change the intended API endpoints of an app board	Read + write	Yes
`author_pages`	Add or change the intended pages of an app board	Read + write	Yes
`author_channels`	Add or change the intended event channels of an app board	Read + write	Yes
`author_boundary_rules`	Add or change the boundary rules over a board's nodes	Read + write	Yes

Insights

Tool	What it does	Scope	Discard undoes it
`list_insight_skills`	The insight templates in your organization	Read	—
`get_insight_skill`	One template's full method	Read	—
`list_insights`	A board's batches of insights	Read	—
`get_insight`	One insight with its evidence and trail	Read	—
`create_insight`	Record a batch of insights on a board	Read + write	No
`promote_insights`	Turn reviewed insights into draft work items	Read + write	No

Work items and epics

Tool	What it does	Scope	Discard undoes it
`list_work_items`	A board's work items	Read	—
`get_work_item`	One work item in full	Read	—
`list_epics`	The workspace's epics	Read	—
`create_work_item`	Draft a work item on a code-bound board	Read + write	No
`update_work_item`	Revise a draft or needs-clarification work item	Read + write	No
`transition_work_item`	Move a work item through its lifecycle	Read + write	No
`assign_work_item`	Assign a work item, or clear its assignees	Read + write	No
`set_work_item_epic`	File a work item under an epic, or take it out	Read + write	No
`delete_work_item`	Delete a work item	Read + write	No
`create_epic`	Create an epic	Read + write	No

Apps, sources and skills

Tool	What it does	Scope	Discard undoes it
`list_source_bindings`	The repos and documents that feed a board	Read	—
`get_source_content`	Read a bound source document	Read	—
`get_skill_profile`	An app board's type and active skill modules	Read	—
`list_skill_library`	The skill modules an app board can activate	Read	—
`convert_node_to_app`	Make a landscape node a repo-bound app	Read + write	No
`bind_reference_source`	Attach a document to a board as reference context	Read + write	No
`configure_skills`	Turn an app board's skill modules on or off	Read + write	No

Connected documents

Tool	What it does	Scope	Discard undoes it
`search_connected_documents`	Find a Google Docs, Confluence or Notion page, or a Jira issue	Read	—
`read_connected_document`	Read one by its link, or by the id a search returned	Read	—

Both use the personal connections of the person who generated the token, so the agent sees only what that person's accounts show.

Hub, publishing and playbooks

Tool	What it does	Scope	Discard undoes it
`get_hub_status`	A board's work item counts, latest insights and source health	Read	—
`list_playbooks`	The playbooks you can start, and the live runs	Read	—
`get_playbook`	One playbook's definition	Read	—
`get_playbook_run`	A run's steps and what proved each one	Read	—
`get_playbook_skill`	A playbook's instructions, for an agent that runs it	Read	—
`publish_board`	Publish a board so people outside the workspace can read it	Read + write	No
`start_playbook`	Start a playbook run	Read + write	No
`report_playbook_step`	Record what the board can't observe on a run you started	Read + write	No
`draft_playbook`	Save a new playbook, disabled	Read + write	No

publish_board makes the same checks as Publish: you own the board, your role can publish, and Protected needs the Pro plan. draft_playbook needs a plan with custom playbooks (Pro). get_hub_status, given a playbook run, counts as that run's visit to the hub.

Working copy

Tool	What it does	Scope	Discard undoes it
`get_write_session`	What's currently uncommitted	Read + write	—
`preview_write_session_commit`	What committing would produce, before doing it	Read + write	—
`discard_write_session`	Undo the whole working copy, row by row	Read + write	—
`commit_write_session`	Accept the working copy and clear the undo log	Read + write	—

You can also accept or discard a token's working copy in the portal, under Plugin pushes awaiting commit. After a commit, nothing in it can be undone. See Review what an agent changed.

What no token gets

The tool set comes from the same registry as Chat, filtered by policy. The policy never exposes tools that need a person in the conversation, or tools that provision new infrastructure. That is why an agent can't post to Slack.