A Lightning Fast CLI for Exa AI Search
Search the web, extract clean markdown content, find similar pages, and generate cited answers directly from your terminal or AI workflows.
Core Advantages
1–3 ms Startup
A single static binary that executes in milliseconds with zero runtime or managed framework overhead.
OS Keystore Security
API keys are securely held in macOS Keychain, Windows DPAPI, or Linux Secret Service, never plain files.
Agentic Protocol
Clean YAML stdout by default, raw JSON via --json, and deterministic machine-readable error codes.
Multi-Account Safety
Manage separate work, personal, and CI accounts with explicit --account targeting.
$ exa search "frontier AI models comparison" --type deep --summary
results:
- title: Frontier AI Models 2026 Comprehensive Benchmark
url: https://arxiv.org/abs/2602.04128
publishedDate: 2026-02-14T08:00:00Z
summary: A multi-benchmark evaluation comparing reasoning latency and context window accuracy...
$ exa answer "How does transformer attention work?" --text
answer: Transformer attention computes weighted representations using queries, keys, and values...
citations:
- https://arxiv.org/abs/1706.03762
Quickstart Workflow
Get started with Exa search in less than two minutes.
Install
Install the CLI using Cargo or download prebuilt binaries from the GitHub releases page.
Login
Run exa login to open the Exa dashboard, copy your API key, and store it in your OS keystore.
Search & Extract
Search with natural language queries, extract markdown contents, or generate answers.
v1.0.0 Stable Release
Download precompiled standalone binaries for macOS, Linux, and Windows.
View full changelog and SHA256 checksums on GitHub Releases.
Command Reference
Organized by operation.
# Natural language web search exa search "latest developments in quantum computing" # Filter by category and highlights exa search "senior ML engineers at fintech companies" --category people --highlights # Filter by news and date range exa search "AI regulation updates" --category news --start-date 2025-01-01 --num 20 # Domain filtering and full page markdown text exa search "react state management" --include-domains "github.com,dev.to" --text # Deep search with LLM summary exa search "frontier AI models comparison" --type deep --summary
# Extract clean markdown text from URL exa contents "https://arxiv.org/abs/2301.07041" --text # Extract query-relevant highlights exa contents "https://example.com/article" --highlights --highlights-query "methodology" # Extract summary and crawl subpages exa contents "https://stripe.com" --summary --subpages 5 --subpage-target "about,careers,press"
# Question answering with live citations exa answer "What is the latest valuation of SpaceX?" # Question answering with full citation text exa answer "How does transformer attention work?" --text
# Find pages similar to an arXiv paper exa find-similar "https://arxiv.org/abs/2307.06435" --num 5 # Find similar blogs, excluding original domain exa find-similar "https://example.com/blog" --highlights --exclude-domains "example.com"
# Interactive login (opens browser, saves as 'default') exa login # Log in to a specific account name exa login work # Pipe API key via stdin (no shell history) printf %s "$KEY" | exa accounts add work --api-key-stdin # List configured accounts and test validity exa accounts list --check # Test connectivity for a stored account exa accounts test work # Remove an account exa accounts remove work --yes