Exa Search API

Name: Exa Search API API
Brand: Exa Search API
Availability: InStock

★ Only Publicly Available OpenAPI DocumentAI/MLEmbeddings SearchapiKey3 EndpointsREST

For Agents

Run neural or keyword web search, retrieve clean page contents, and find similar URLs through three Exa endpoints designed for AI agents.

Quickstart

Get started with Exa Search API in minutes using your preferred integration method.

# Add to your MCP client config (Claude Desktop, Cursor, Windsurf)
{
  "jentic": {
    "url": "https://api.jentic.com/mcp",
    "auth": "oauth"
  }
}

# Then ask your agent:
"search the web with exa"

# → Jentic returns the GET /events tool with parameter schema, agent executes.

Capabilities

What an agent can do with Exa Search API API.

Run a neural or keyword search via POST /search and receive ranked URLs

Fetch clean text contents for a list of URLs via POST /contents

Discover semantically similar URLs to a given source via POST /findSimilar

Compose search-then-fetch in two calls to ground LLM responses on live web text

GET STARTED

Start building with Exa Search API API

Explore with Jentic

View OpenAPI Document

Jentic publishes the only available OpenAPI document for Exa Search API, keeping it validated and agent-ready.

Jentic publishes the only available OpenAPI specification for Exa Search API, keeping it validated and agent-ready. Exa is an AI-powered search engine purpose-built for AI agents and applications. The API offers neural and keyword search, structured page-content retrieval, and similar-link discovery across just three endpoints, making it the cleanest possible search primitive for an LLM agent. Exa understands natural-language queries and returns ranked URLs with optional clean text contents instead of raw HTML.

Use Cases

Patterns agents use Exa Search API API for, with concrete tasks.

★ Grounded LLM Search-and-Synthesise

RAG-style assistants need fresh web text to answer questions about current events. Exa's POST /search returns ranked URLs and POST /contents returns clean body text with HTML stripped, so an agent can grab the top five results, fetch their contents, and feed them into a language model in two HTTPS calls. The cleaned-up text avoids the boilerplate and ad markup that pollute raw scraping.

POST /search with the user query and num_results=5, then POST /contents with the returned URL list and pass the cleaned text into the LLM

Competitor and Citation Discovery

Researchers, marketers, and journalists need to find pages similar to a known reference — competitor blog posts, similar academic abstracts, or news takes on the same story. POST /findSimilar takes one URL and returns semantically related URLs, surfacing competitive content or alternate sources without keyword fiddling.

POST /findSimilar with the URL of a published article to retrieve 10 semantically similar pages for citation research

Curated Domain or Date-Filtered Search

Some workflows only want results from specific domains (e.g. .edu sites, a list of approved news outlets) or within a recent date window. Exa's /search payload accepts include_domains, exclude_domains, and date filters, letting an agent constrain results to a curated subset and avoid noisy general-web answers.

POST /search with query='climate policy 2026', include_domains=['nature.com','science.org'], start_published_date='2026-01-01' to get curated recent results

AI Agent Web Search via Jentic

Agentic frameworks (LangChain, CrewAI, custom MCP clients) need a single search primitive that returns clean text. Jentic exposes Exa's three endpoints as MCP tools so an agent searches by intent ('search the web for X'), runs POST /search, optionally fetches contents, and synthesises an answer — all without writing a custom Exa client.

Use Jentic search 'search the web with exa' to find POST /search, load schema, and execute with the user's query, then chain into POST /contents for the top three URLs

Key Endpoints

3 endpoints — jentic publishes the only available openapi specification for exa search api, keeping it validated and agent-ready.

METHOD

PATH

DESCRIPTION

POST

/search

Run neural or keyword web search

POST

/contents

Fetch clean text for a list of URLs

POST

/findSimilar

Find URLs similar to a given source

POST

/search

Run neural or keyword web search

POST

/contents

Fetch clean text for a list of URLs

POST

/findSimilar

Find URLs similar to a given source

Why though Jentic?

Three things that make agents converge on Jentic-routed access.

Credential isolation

The Exa apiKey is stored encrypted in the Jentic MAXsystem vault. Agents receive scoped execution rights only — Jentic injects the key per call so it never appears in agent prompts or logs.

Intent-based discovery

Agents search by intent (e.g. 'search the web with exa') and Jentic returns POST /search with its full query-payload schema, so the agent can pass include_domains, date filters, and num_results without reading vendor docs.

Time to first call

Direct Exa integration: a few hours to wire search and contents calls. Through Jentic: minutes — search, load schema, execute, with credentials handled out of band.

Related APIs

Alternatives and complements available in the Jentic catalogue.

Alternative

Tavily Search API

Tavily is another AI-first web search API geared towards RAG agents.

Choose Tavily when you need built-in summarisation alongside search; choose Exa for cleaner neural ranking and similarity discovery.

Alternative

Perplexity API

Perplexity returns answers with citations rather than raw URL lists.

Choose Perplexity when the agent needs a synthesised answer with citations; choose Exa when it needs raw URLs and clean text to compose its own answer.

Complementary

Anthropic API

Claude language models that can ingest the cleaned text Exa returns.

Pair Exa with Anthropic when building a RAG flow where Claude synthesises answers from Exa-fetched contents.

FAQs

Specific to using Exa Search API API through Jentic.

Why is there no official OpenAPI spec for Exa Search API?

Exa does not publish an OpenAPI specification. Jentic generates and maintains this spec so that AI agents and developers can call Exa Search API via structured tooling. It is validated against the live API and kept up to date. Get started at https://app.jentic.com/sign-up.

What authentication does the Exa Search API use?

The API uses an apiKey credential carried over HTTP. Through Jentic, the key sits encrypted in the MAXsystem vault and is injected at call time so it never appears in agent context.

Can I get clean page text along with search results from Exa?

Yes — call POST /search to get ranked URLs, then POST /contents with that URL list to receive cleaned body text with boilerplate removed. The two-call pattern is the standard RAG flow with Exa.

What are the rate limits for the Exa Search API?

The OpenAPI spec does not declare quantitative rate limits; Exa applies plan-based throttling. Inspect 429 responses and Retry-After headers in production and confirm your plan ceiling with Exa support.

How do I run an Exa search through Jentic?

Run pip install jentic, then await client.search('search the web with exa'), client.load, client.execute. Jentic returns POST /search with the query payload schema wired in for the agent to fill.

How is /findSimilar different from /search on the Exa Search API?

POST /search takes a natural-language query and returns matching URLs, while POST /findSimilar takes one URL and returns semantically similar URLs. Use /search for fresh queries; use /findSimilar to find competitors or alternate sources to a known reference.