hm_find_related

Find nodes connected to a given node by traversing hyperedge relationships. The server picks a traversal strategy from your query and optional pattern hints.

Parameters

Parameter	Type	Default	Description
`start_node`	`str`	required	Starting node key (must exist in the graph)
`query`	`str \| None`	`None`	Natural-language hint; influences strategy selection
`max_nodes`	`int`	`50`	Cap on nodes returned (max 500)
`max_depth`	`int \| None`	`None`	Depth bound for decay-style traversal (max 10)
`relationship_pattern`	`str \| None`	`None`	Optional pattern hint for chain traversal
`session_id`	`str`	`"default"`	Session scope for ranking context
`lens`	`str \| None`	`None`	Optional lens filter label

Traversal strategies

The server supports three traversal strategies:

Strategy	Description
Chain	Follow edges step-by-step from the starting node
Decay	Weight connections by distance — closer nodes score higher
Type-boosted	Prioritize specific node types in traversal

Response

Returns the subgraph reachable from the starting node — connected nodes with their relationships and metadata.

Example

{
  "start_node": "person_alice",
  "query": "who worked on the launch",
  "max_nodes": 50
}

CLI equivalent

hm find person_alice --depth 2

REST endpoint

POST /api/v1/memory/find-related

Pass the same fields as JSON (e.g. start_node, not key).