web2-recon

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides a wide array of shell commands to automate security testing processes, including subdomain discovery, port scanning, and historical URL retrieval. It also includes instructions to modify the user's shell profile (~/.zshrc) to persist environment variables for API keys.- [EXTERNAL_DOWNLOADS]: The skill installs several third-party security tools via pip and pipx, such as trufflehog3, git-dumper, and ds-store, and downloads updated vulnerability templates for the nuclei scanning engine.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of fetching and analyzing untrusted content from external websites (JavaScript bundles, HTML pages, and server configuration files).
  • Ingestion points: The skill ingests data from external targets using tools like katana, waybackurls, and curl.
  • Boundary markers: No specific delimiters are used to isolate external data from the command execution environment.
  • Capability inventory: The skill utilizes shell execution, network access, and file system writes to manage recon output and configuration.
  • Sanitization: No sanitization of ingested content is performed before analysis.- [CREDENTIALS_UNSAFE]: The setup instructions direct the user to store sensitive API keys in plaintext within local configuration files and shell environment scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 05:47 PM
Security Audit — agent-trust-hub — web2-recon