Semantic Search

Vector search across entities.

1 action. Each is callable by the agent as a tool, over HTTP via /api/invoke, and (usually) from a UI page.

search.global

POST

Unified fuzzy + semantic search across ALL entity types in the current workspace (subjects, contacts, tickets, drafts, knowledge, skills, missions, projects, documents, conversations, engagements, decisions, actors). Returns one ranked list of lightweight hits with deep links — results are RBAC-scoped to what the caller may see. Use this for a global "find anything" box; use entity.search when you already know the kind and want full typed rows.

When the agent calls it: Default global finder. mode="hybrid" blends full-text, fuzzy (typo-tolerant), and semantic ranking. Pass types to scope, e.g. types=["contact"]. Each hit carries a url you can navigate to.

• AI tool: search_global
• API: POST /api/invoke/search.global

Input

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "type": "object",
  "properties": {
    "q": {
      "type": "string",
      "minLength": 1
    },
    "types": {
      "type": "array",
      "items": {
        "type": "string"
      }
    },
    "mode": {
      "type": "string",
      "enum": [
        "keyword",
        "semantic",
        "hybrid"
      ]
    },
    "limit": {
      "type": "integer",
      "minimum": 1,
      "maximum": 50
    },
    "offset": {
      "type": "integer",
      "minimum": 0,
      "maximum": 9007199254740991
    }
  },
  "required": [
    "q"
  ]
}