agent-cve-lookup

Queries candidate CVEs for observed services using NVD and CISA KEV data.

Updated May 27, 2026
One-click install
npx skills add https://github.com/baljinnyamday/autonomous-red-team-agent --skill agent-cve-lookup-baljinnyamday
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-cve-lookup
Source: https://github.com/baljinnyamday/autonomous-red-team-agent/tree/main/agent-baseline/.agents/skills/agent-cve-lookup
Command: npx skills add https://github.com/baljinnyamday/autonomous-red-team-agent --skill agent-cve-lookup-baljinnyamday

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? During an authorized red team engagement, an agent discovers versioned services but lacks vulnerability hypotheses for them. This Skill looks up candidate CVEs for observed services so the agent can prioritize investigation without guessing or querying hidden topology data. ## Core Features & Use Cases - Service-Based CVE Search: Runs agent-baseline cve-search against observed hosts or services recorded in the world model, using product and version hints from services.md. - Prioritized Results: Exports candidates to vulns.md ranked by CISA KEV membership, CVSS severity, confidence, and CPE versus keyword-only matches. - Official Data Sources: Matches against the NVD CPE API, NVD CVE API, and CISA KEV JSON, with optional cache refresh for stale source data. - Use Case: After reconnaissance reveals Apache httpd 2.4.49 on a host, run the lookup, export the vault, and read the host's vulns.md to get verified-source vulnerability candidates for planning. ## Quick Start Ask the agent to look up CVE candidates for the services observed on a specific host in the current run and show the prioritized results from that host's vulns.md file.

Frequently Asked Questions about agent-cve-lookup

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I look up CVEs for a discovered service during a pentest?

Run `agent-baseline cve-search` with the host or service identifier against the run's world.sqlite database, then export the vault and read the host's vulns.md file. The results list candidate CVEs matched from NVD and CISA KEV sources.

What inputs give good CVE lookup results?

Strong inputs include a product name plus version, such as OpenSSH 8.2p1 or Apache httpd 2.4.49. Service names alone or missing versions produce lower confidence and noisier keyword-based matches.

How are CVE lookup results prioritized?

Candidates are prioritized by CISA KEV membership first, then CVSS severity, then confidence score, with CPE-based matches ranked above keyword-only matches. This ordering appears in the exported vulns.md file.

Can CVE candidates be used directly as exploit instructions?

No. Results are hypotheses only and must be verified against the exact product, version, and configuration before use. The skill explicitly prohibits generating exploit commands from lookup output alone.

What data sources does the CVE search use?

The tool queries the NVD CPE API for product/version matching, the NVD CVE API for candidate vulnerabilities, and the CISA KEV JSON feed for known-exploited enrichment. A --refresh-cache flag updates stale cached source data.