> ## Documentation Index
> Fetch the complete documentation index at: https://docs.secapi.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# GET /v1/search/semantic

> Semantic vector search across SEC filing section content with finance-tuned embeddings, hybrid keyword + vector RRF, and citation fields on every result row

Semantic vector search across SEC filing section content with finance-tuned embeddings, hybrid keyword + vector RRF, and citation fields on every result row

<Info>
  Audience: application and coding agent.
</Info>

## Citation fields

Every result row carries seven additive top-level fields:

| Field                     | Description                                                                                                                      |
| ------------------------- | -------------------------------------------------------------------------------------------------------------------------------- |
| `accession`               | SEC accession number for the source filing                                                                                       |
| `section_key`             | Canonical section identifier (e.g. `item_1a`)                                                                                    |
| `char_start` / `char_end` | Half-open offsets into the section markdown text (`section_snippets.content_md`) — section-text-relative, stable across reparses |
| `highlighted_snippet`     | ±150 char window with query terms wrapped in `**…**`, sentence-boundary truncated, ≤320 chars                                    |
| `source_url`              | Public SEC.gov filing URL                                                                                                        |
| `ticker`                  | Issuer ticker, when known                                                                                                        |

When the validator can't produce a span (no section text, no query-term match, or offsets out of bounds), `char_start`/`char_end`/`highlighted_snippet` drop to `null` and `_citation_degraded` records the reason. Identity fields (`accession`, `section_key`, `source_url`, `ticker`) are always populated when known. See [api conventions → citation + char-range spans](/api-conventions#citation-char-range-spans) for the full contract.

## Agent mode

Pass `?view=agent` to drop `score` and `retrievalMode` from each row. The seven citation fields are preserved unconditionally. See [api conventions → response formats](/api-conventions) for the full `?view=` contract.

## Canonical metadata

* `requestId`
* `traceparent`

## Example request

<RequestExample>
  ```bash theme={null}
  curl -X GET \
    -H "x-api-key: $SECAPI_API_KEY" \
    -H "secapi-version: 2026-03-19" \
    "https://api.secapi.ai/v1/search/semantic?q=supply%20chain%20risk&ticker=AAPL&form=10-K&mode=hybrid&view=agent"
  ```
</RequestExample>

## Example response

<ResponseExample>
  ```json theme={null}
  {
    "requestId": "req_2ZK8Q1W9F4M6P7R3"
  }
  ```
</ResponseExample>

## Give this prompt to your agent

<Prompt>
  Use SEC API GET /v1/search/semantic to semantic vector search across SEC filing section content with finance-tuned embeddings, hybrid keyword + vector RRF, and citation fields on every result row. Preserve `requestId`, `traceparent` in the output. Return a concise markdown summary plus the exact structured payload fields that a downstream engineer or agent should keep using this result.
</Prompt>

## Failure posture

* treat non-2xx responses as contract-aware failures, not free-form errors
* preserve `requestId` and `traceparent` in logs and downstream reports
* if provenance or freshness metadata is present, return it unchanged so trust is not lost in the handoff