# exa

> Exa CLI tool for intelligent web search, content extraction, and code context retrieval.

- Author: 9bingyin
- Repository: 9bingyin/skills
- Version: 20260123051502
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-06
- Source: https://github.com/9bingyin/skills
- Web: https://mule.run/skillshub/@@9bingyin/skills~exa:20260123051502

---

---
name: exa
description: "Exa CLI tool for intelligent web search, content extraction, and code context retrieval."
---

# Exa CLI

Exa is a search engine built from the ground up for AIs. This tool provides a zero-dependency Python CLI to access Exa's powerful search, content, and code context capabilities.

## Usage

### 1. Search
Intelligently search the web and extract contents. Exa automatically chooses the best search method using an embeddings-based model.

**Command:**
```bash
python3 scripts/exa_cli.py search "QUERY" [OPTIONS]
```

**Core Options:**
- `query` (required): The search query string.
- `--num-results <int>`: Number of results (1-100). Default: 10.
- `--type <neural|auto|fast|deep>`: 
    - `neural`: Semantic search using embeddings.
    - `auto` (default): Intelligently combines neural and other search methods.
    - `fast`: Streamlined version of search models for speed.
    - `deep`: Comprehensive search with query expansion and detailed context.
- `--additional-queries <list>`: Additional query variations for `deep` search.

**Filtering Options:**
- `--category <string>`: Focus on data categories: `company`, `research paper`, `news`, `pdf`, `github`, `tweet`, `personal site`, `financial report`, `people`.
- `--include-domains <list>`: Only include results from these domains.
- `--exclude-domains <list>`: Exclude results from these domains.
- `--user-location <string>`: Two-letter ISO country code (e.g., `US`).
- `--start-crawl-date <iso>` / `--end-crawl-date <iso>`: Filter by when Exa discovered the link.
- `--start-published-date <iso>` / `--end-published-date <iso>`: Filter by publication date.
- `--include-text <list>`: Strings that *must* be present in the webpage text (max 5 words).
- `--exclude-text <list>`: Strings that *must not* be present in the text.
- `--moderation`: Enable content moderation.

**Content Options:**
- `--text`: Return full page text.
- `--summary`: Return AI-generated page summary.
- `--highlights`: Return relevant text highlights identified by LLM.

**Example:**
```bash
python3 scripts/exa_cli.py search "latest LLM research" --category "research paper" --num-results 5 --summary --start-published-date "2024-01-01T00:00:00.000Z"
```

### 2. Contents
Get full page contents, summaries, and metadata for specific URLs. Returns instant results from cache with automatic live crawling as fallback.

**Command:**
```bash
python3 scripts/exa_cli.py contents URL [URL...] [OPTIONS]
```

**Core Options:**
- `urls` (required): List of URLs to retrieve.
- `--livecrawl <never|fallback|preferred|always>`:
    - `never`: Disable livecrawling (cache only).
    - `fallback` (default): Livecrawl when cache is empty.
    - `preferred`: Always try to livecrawl, fall back to cache if it fails.
    - `always`: Always live-crawl, never use cache.
- `--livecrawl-timeout <int>`: Timeout for livecrawling in milliseconds.
- `--subpages <int>`: Number of subpages to crawl.
- `--subpage-target <list>`: Terms to find specific subpages (e.g., `pricing`).

**Content Options:**
- `--text`: Return full page text.
- `--summary`: Return page summary.
- `--highlights`: Return relevant highlights from the page.

**Example:**
```bash
python3 scripts/exa_cli.py contents "https://exa.ai/blog/exa-code" --text --livecrawl preferred --subpages 1
```

### 3. Context (Exa Code)
Get relevant code snippets and examples from open source libraries and repositories. Searches billions of GitHub repos and docs to find token-efficient context for coding.

**Command:**
```bash
python3 scripts/exa_cli.py context "QUERY" [OPTIONS]
```

**Options:**
- `query` (required): Search query to find relevant code snippets (e.g., "how to use react hooks").
- `--tokens <int|dynamic>`: Token limit for response. `dynamic` (default) determines optimal length. Specific values (e.g., 5000) allow precise control.

**Example:**
```bash
python3 scripts/exa_cli.py context "React hooks for state management" --tokens 5000
```

## Pricing

**COST OPTIMIZATION:** Always strive to minimize costs. Avoid adding unnecessary search parameters (like `--text` or `--summary`) or requesting more results than strictly needed (keep `--num-results` low unless necessary). Costs depend on search type, number of results, and content extraction.

### 1. Search Costs
Base cost per request based on `--type` and `--num-results`:

| Type | Results | Cost |
| --- | --- | --- |
| Neural/Auto/Fast | 1-25 | $0.005 |
| Neural/Auto/Fast | 26-100 | $0.025 |
| Neural/Auto/Fast | 100+ | $1.00 |
| Deep | 1-25 | $0.015 |
| Deep | 26-100 | $0.075 |

**Content Extractions:**
Adding `--text`, `--summary`, or `--highlights` incurs extra "Contents" costs (see below).

### 2. Contents Costs
Charged per 1,000 tokens (approx. 1 "page") of processed content.

| Feature | Cost per 1k Tokens |
| --- | --- |
| Text (`--text`) | $0.001 |
| Highlights (`--highlights`) | $0.001 |
| Summary (`--summary`) | $0.001 |

### 3. Context (Code) Costs
- **Search Fee:** ~$0.005 per request
- **Content Fee:** $1.00 per 1M tokens (approx $0.001 per 1k tokens) based on returned length.

*Tip: Check the `costDollars` field in the API response for exact usage.*

## Help
```bash
python3 scripts/exa_cli.py --help
python3 scripts/exa_cli.py <command> --help
```