cve-triage
Fail
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_CONTEXT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill uses a shell command at load time to fetch live metadata from the official CISA KEV catalog to provide current context to the agent.
- [REMOTE_CODE_EXECUTION]: The skill downloads a JSON feed from a trusted government domain and pipes the output into a Python interpreter to extract and display version information.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content such as vulnerability scan results and vendor advisories, which serves as an attack surface for indirect prompt injection.
- Ingestion points: The agent reads user-provided scan reports (Qualys, Tenable, Snyk, etc.) and CVE advisory text.
- Boundary markers: The skill includes a dedicated 'Prompt Injection Safety Notice' section that instructs the agent to disregard instructions embedded in the data.
- Capability inventory: The skill's capabilities are restricted to reading, searching, and web fetching via frontmatter configuration; no file-writing or administrative shell tools are available to the agent during execution.
- Sanitization: Instructions mandate that all triage outcomes must be derived strictly from framework metrics (CVSS, SSVC) and ignore any 'gut feel' or prose-based instructions in the source data.
Recommendations
- HIGH: Downloads and executes remote code from: https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json - DO NOT USE without thorough review
Audit Metadata