Search

HelperIQ provides a unified search endpoint plus per-resource search endpoints. All search is full-text (PostgreSQL trigram + tsvector indexes — no Elastic dependency).

Method	Path	Permission
GET	/api/v1/search	conversations:read
GET	/api/v1/conversations/search	conversations:read
GET	/api/v1/messages/search	messages:read
GET	/api/v1/contacts/search	contacts:read

Unified search

GET /api/v1/search?q=<query>

Searches across conversations, messages, and contacts simultaneously. Used by the global Ctrl+K search palette in the agent UI.

Query parameters:

Param	Default	Description
q	— required	Search query. Minimum 2 characters.
limit	10 per category	Max results per category.

Response:

{
  "data": {
    "conversations": [
      {
        "uuid": "afa55a3b-…",
        "reference_number": 217,
        "subject": "Login problem after password reset",
        "snippet": "…can't <mark>login</mark> after I reset my password…",
        "contact": { "first_name": "Linda", "email": "[email protected]" }
      }
    ],
    "messages": [
      {
        "uuid": "b29d68b1-…",
        "conversation_uuid": "afa55a3b-…",
        "snippet": "…tried to <mark>login</mark> three times…",
        "created_at": "…"
      }
    ],
    "contacts": [
      { "id": 42, "first_name": "Linda", "last_name": "Olsen", "email": "[email protected]" }
    ]
  }
}

Matches include a snippet with the matched query highlighted via <mark> tags — convenient to render directly in HTML.

Search conversations

GET /api/v1/conversations/search?q=<query>

Returns conversations whose subject or first/last message matches.

Query parameters:

Param	Default	Description
q	— required	Query. Min 2 chars.
page, page_size	1, 50	Standard pagination.
filters	—	JSON-encoded ad-hoc filter array (see API Introduction → Filtering). Combines with the text query.

Response shape matches the conversation list endpoints.

Search messages

GET /api/v1/messages/search?q=<query>

Full-text search over message content (HTML stripped). Useful for finding "the email that mentioned the SKU X42".

Query parameters: q, page, page_size, optional conversation_uuid to scope to a single thread.

Response:

{
  "data": {
    "results": [
      {
        "uuid": "b29d68b1-…",
        "conversation_uuid": "afa55a3b-…",
        "snippet": "…order <mark>X42</mark> was shipped on…",
        "type": "outgoing",
        "private": false,
        "created_at": "…"
      }
    ],
    "total": 12, "total_pages": 1, "page": 1, "page_size": 50
  }
}

Search contacts

GET /api/v1/contacts/search?q=<query>

Quick lookup over contact name + email + phone. Used by the recipient autocomplete in the reply editor and the "+ New conversation" form.

Returns up to 20 matches; no pagination.

{
  "data": [
    { "id": 42, "first_name": "Linda", "last_name": "Olsen", "email": "[email protected]", "phone": "…" }
  ]
}

Search index lifecycle

The search indexes are PostgreSQL GIN indexes maintained automatically:

• idx_trgm_conversation_messages_on_text_content (trigram on plain text)
• Per-table tsvector indexes on subject, contact name, etc.

Indexes are updated synchronously on insert/update — no separate ingestion process. New conversations are searchable immediately after the writing transaction commits.