# Exa API

Proxies the Exa API. Ephemeral tenants get the Exa core retrieval surface, Claimed Free adds research plus deep search and deep-reasoning search, and Plus/Pro unlock websets and monitors. No Exa API key is needed.

## Prerequisites

Edge calls require this resource to be provisioned first; unprovisioned calls return an error.

### Provision

    curl -s -X POST https://cohesivity.ai/api/resources/exa-api \
      -H "Authorization: Bearer <coh_management_key>"

### Delete

    curl -s -X DELETE https://cohesivity.ai/api/resources/exa-api \
      -H "Authorization: Bearer <coh_management_key>"

Provisioning happens once, before the application runs; the running application does not provision its own resources.

## Official Docs

https://exa.ai/docs/reference covers full request and response details.

> **Server-side only.** `coh_application_key` is a secret; browser JS, mobile bundles, and other client-side code cannot hold it safely. This call belongs in a Railway-hosted server, `cloudflare-workers`, or your own server tier. See the canonical key-secrecy directive in `.cohesivity` for details.

## Edge Usage

- **Base URL:** https://cohesivity.ai/edge/exa-api
- **Auth:** `coh_application_key` as the **key** query parameter
- **Method:** mostly POST with JSON body; Plus/Pro may also use Exa's webset and monitor endpoints through the same proxy
- **Format:** matches Exa's API: same paths, same request/response shapes

## Launch Access Policy

- **Ephemeral core surface:** `POST /search`, `POST /contents`, `POST /findSimilar`, `POST /answer`, and `POST /context`
- **Ephemeral search-type restriction:** `/search` only allows `instant`, `fast`, `auto`, and `neural`
- **Blocked on Ephemeral:** `/research*`, `/websets*`, `/monitors*`, and `/search` types `deep-lite`, `deep`, `deep-reasoning`
- **Claimed Free:** core retrieval, `/research*`, and deep search types including `deep-lite`, `deep`, and `deep-reasoning`; websets and monitors require Plus or Pro
- **Claimed Plus and Claimed Pro:** claimed search types plus websets and monitors

## Products

- **Search** (`POST /search`): find web pages matching a query. Claimed tenants can use deep search types, including `deep-reasoning`, through the same endpoint. Returns titles, URLs, dates, and optionally text/summaries.
- **Contents** (`POST /contents`): extract text, highlights, or summaries from a list of URLs.
- **Find Similar** (`POST /findSimilar`): find pages similar to a URL or document.
- **Context** (`POST /context`): return compact context payloads for downstream reasoning.
- **Answer** (`POST /answer`): ask a question and get the final synthesized answer with citations. This is the LLM answer surface, not just raw retrieval.
- **Research** (`/research*`): claimed tenants can use Exa research surfaces. Ephemeral tenants must claim first.
- **Monitors** (`/monitors*`): watch topics or pages over time. Plus/Pro only.
- **Websets** (`/websets/v0/websets*`): build structured datasets from web data asynchronously. Plus/Pro only.

## Examples

- Search: `POST https://cohesivity.ai/edge/exa-api/search?key=<coh_application_key>` with body `{ "query": "...", "numResults": 10, "type": "auto" }`
- Contents: `POST https://cohesivity.ai/edge/exa-api/contents?key=<coh_application_key>` with body `{ "urls": ["https://example.com"], "text": true }`
- Answer: `POST https://cohesivity.ai/edge/exa-api/answer?key=<coh_application_key>` with body `{ "query": "..." }`
- Context: `POST https://cohesivity.ai/edge/exa-api/context?key=<coh_application_key>` with body `{ "query": "..." }`
- Research (claimed tiers): `POST https://cohesivity.ai/edge/exa-api/research?key=<coh_application_key>` with the body shape from Exa's current docs
- Deep reasoning search (claimed tiers): `POST https://cohesivity.ai/edge/exa-api/search?key=<coh_application_key>` with body `{ "query": "...", "numResults": 1, "type": "deep-reasoning" }`
- Websets (Plus/Pro): `POST https://cohesivity.ai/edge/exa-api/websets/v0/websets?key=<coh_application_key>` with body `{ "search": { "query": "...", "count": 10 }, "enrichments": [{ "description": "...", "format": "text" }] }`

## Response Format

Search and deep-reasoning search: `{ results: [{ title, url, publishedDate?, author?, text?, summary? }], requestId?, autopromptString? }`

Contents: `{ results: [{ title, url, text, highlights?, summary? }] }`

Answer: `{ answer: "...", citations: [{ url, title, text? }] }`

Websets: `{ id: "...", status: "...", dashboardUrl: "..." }` (async: results require polling)

## Billing and Usage

- Claimed Exa usage is fluid-only. Successful usage is billed at provider cost plus 10%, rounded up to the nearest cent per settled charge.
- Failed upstream responses do not burn quota or fluid; Cohesivity revokes the preflight counters before returning the failure.

## Rate Limits

Ephemeral tenants pause as a whole if any authoritative hard cap below is exceeded. Claimed tiers use account-scoped buckets shared across every project owned by the Cohesivity user; OpenAI, AI Gateway, Deepgram, and Exa are fluid-only after tier, rate, and concurrency checks; AI Gateway and Deepgram have no fixed monthly usage bucket for claimed tiers.

**Ephemeral**

- requests: 50 per ephemeral tenant lifetime before claim or expiry
- requests: 5 per minute

**Claimed Free**

- requests: 15 per minute

**Claimed Plus**

- requests: 60 per minute

**Claimed Pro**

- requests: 300 per minute

### Notes

- Exa is fluid-only for claimed accounts. Ephemeral tenants get core retrieval only, Claimed Free adds research plus deep search and deep-reasoning search, and Plus/Pro unlock websets and monitors.