Skip to main content
Searches the knowledge graph using vector similarity, optionally traversing graph connections.

​
Parameters

​
query
string
required
Search query text.
​
limit
number
default:"10"
Maximum number of results (1-100).
​
traverseDepth
number
default:"1"
Graph traversal depth from matched nodes.
ValueBehavior
0No traversal — only vector search hits
1Include direct neighbors (default)
2Include neighbors of neighbors (broader but noisier)
​
scope
string
default:"project"
Search scope.
ValueDescription
projectCurrent project + global (default)
globalGlobal only
allAll projects without filtering

​
Search Strategy

  1. Vector search on entity names → direct hits (score: 1.0)
  2. Graph traversal from hit nodes → connected entities (score: 0.5)
  3. Vector search on edge facts → matching relationships (score: 0.8)
Results are merged and sorted by score.

​
Example

{
  "query": "frontend framework",
  "limit": 10,
  "traverseDepth": 1
}
Returns nodes matching “frontend framework” plus their 1-hop neighbors and relevant edges.