Drill Tree

Walk a tree-shaped drillable resource at three depths (index|summary|full). Default depth is index — cheapest walk. Today's domains: knowledge (units via agent_docs), summary (summarization framework's per-node store).

MCP name: drill_tree

Category: disclosure

Parameters

Name	Type	Required	Description
depth	str	No	'index' (default; node + child names only — cheapest), 'summary' (node + each child's summary text), or 'full' (everything).
domain	str	Yes	Registered domain name. Today: 'knowledge' or 'summary'. Use available_domains() / inspect the disclosure system unit to see all registered.
node_id	str	Yes	Domain-specific node identifier. knowledge: unit path (e.g. 'architecture/summarization-framework'). summary: '{namespace}:{item_id}' for the whole tree or '{namespace}:{item_id}#n{ordinal}' for an internal node.

Details

Universal tree navigation across registered TreeDrillable domains.

drill_tree and walk are the same op — walk is the canonical short name. Both are stable; use whichever reads naturally in context.

When to reach for this vs. other tools:

• Use drill_tree when you already have a node id and want to navigate its structure. No ranking is performed; the response is the requested view at the chosen depth.
• Use summary_search when you have a topic / query and need to find which items match it. See summarization/summary_search. summary_search's stage-1 hits surface a drill_node_id field that drops straight into drill_tree(domain="summary", ...) with no string translation.
• Use context_search for general IR search across any registered source.

Depths

• index (default) — this node + child names only. Cheapest. Use for catalog browsing or when you only need to know what's available.
• summary — this node + each child's summary text. Use for triage ("which child matters?") without paying for full content.
• full — everything: this node's full content + the subtree concat for tree domains.

Default is index rather than summary so an agent reaching for drill_tree(node_id="{namespace}") gets a cheap catalog list rather than triggering N per-child summary fetches; opt into deeper views explicitly.

Domains today

summary — framework summaries

Node-id shapes: - {namespace} (no colon) — namespace root. Children are every summary item under that namespace, ordered by generated_at DESC. - {namespace}:{item_id} — the whole item (root of one summarized session or page). Children are level-1 topic nodes. - {namespace}:{item_id}#n{ordinal} — a specific node within the tree.

knowledge — knowledge units

Node_id is the unit path (tasks/triage-directions, architecture/summarization-framework, etc.). Wraps agent_docs(path=, depth=) so the same walk-by-id verb covers the knowledge store and the summary store.

Adding a new domain

Implement a class with domain: str and get(node_id, depth) -> TreeView (raise DrillError on bad input), then register via register_drillable(domain, factory). New tree-shaped domains plug in immediately — no per-domain capability to write.