cve-triage
Fail
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill employs dynamic context injection to execute a shell command during the initial loading phase. This command is used to fetch the current version of the CISA Known Exploited Vulnerabilities (KEV) catalog.
- [EXTERNAL_DOWNLOADS]: Retrieves vulnerability metadata from the official Cybersecurity and Infrastructure Security Agency (CISA) website, a well-known government resource.
- [REMOTE_CODE_EXECUTION]: Downloaded JSON content from the CISA feed is piped into a Python process for metadata parsing and version string extraction.
- [PROMPT_INJECTION]: The skill is designed to ingest and process untrusted vulnerability scan results, which presents a surface for indirect prompt injection. Ingestion points: Analyzes external scan outputs and CVE identifiers in Step 1. Boundary markers: The skill includes a 'Prompt Injection Safety Notice' that explicitly instructs the agent to ignore any instructions found within ingested scan data. Capability inventory: The skill utilizes WebFetch and file access tools (Read, Grep, Glob) during the triage process. Sanitization: It relies on instructional guardrails to prevent the agent from following instructions embedded in external vulnerability reports.
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