| name | exa-search |
| description | Semantic search using Exa API for security research queries. Unlike keyword-based search, Exa understands context and retrieves high-quality, relevant results for technical research. |
| origin | openclaw |
| version | 0.2.0.2 |
| compatibility | ["openclaw","claude-code","cursor","windsurf"] |
| allowed-tools | ["Bash","Read","Write","Edit","WebSearch","WebFetch","Agent"] |
| defense_triple_required | false |
| metadata | {"domain":"research","tool_count":0,"guide_count":5,"last_reviewed":"2026-07-26"} |
Exa Search
Summary
Unlike keyword-based search, Exa understands context and retrieves high-quality, relevant results for technical research.
Domain: research
Skill Identity
| Attribute | Value |
|---|
| Domain | Intelligence Gathering |
| Skill ID | exa-search |
| Version | 1.0.0 |
| Hacker Laws | Law 3 (Intelligence Over Force), Law 9 (Systematic Over Random) |
| Related Skills | deep-research, osint, social-intelligence |
Purpose
Semantic search using Exa API for security research queries. Unlike keyword-based search, Exa understands context and retrieves high-quality, relevant results for technical research.
Core Capabilities
- Semantic Search: Context-aware query understanding
- Date Filtering: Recent content prioritization
- Domain Filtering: Target specific sources
- Content Extraction: Full-text retrieval
Use Cases
- CVE Research: "Recent CVEs affecting Spring Boot applications"
- Exploit Techniques: "SSRF bypass techniques in AWS metadata service"
- Tool Research: "Best tools for JWT security testing"
- Threat Intelligence: "APT campaigns targeting healthcare 2024"
Exa API Reference
Query Strategy
- Semantic phrasing: Frame queries as full sentences ("How does X bypass Y?") rather than keyword strings; Exa's embedding model rewards natural-language intent.
- Iterative narrowing: Start broad, then re-query with terms surfaced by the first batch to drill into specifics.
- Source class filters: Restrict to
includeDomains (e.g., ["github.com","arxiv.org"]) for code/research, or excludeDomains for noisy aggregators.
