attack-surface-recon
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses terminal commands to run a suite of security tools including amass, subfinder, nmap, nuclei, ffuf, and katana for network mapping, port scanning, and vulnerability detection.
- [EXTERNAL_DOWNLOADS]: The skill fetches data from multiple external sources including crt.sh, web.archive.org, and public search engines like FOFA and Shodan. It also crawls target web assets to identify potential secrets and endpoints.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from target assets to guide its autonomous reconnaissance logic. Ingestion points: Processes external data from certificate logs (crt.sh), archives, and HTTP responses from target domains. Boundary markers: No explicit delimiters are used to wrap external content when the agent processes it to identify frameworks or versions. Capability inventory: The skill possesses significant capabilities including network scanning, file system writes (tee), and arbitrary terminal execution across various tools. Sanitization: External data is processed via grep and parsed as JSON, but there is no specific sanitization logic to prevent malicious input from influencing the agent's logic flow.
Audit Metadata