audit-recon-links

Pass

Audited by Gen Agent Trust Hub on Sep 8, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches the 'katana' crawling tool from the official Project Discovery GitHub repository using the Go package manager.
  • [EXTERNAL_DOWNLOADS]: Downloads Linux repository signing keys from Google's official infrastructure to enable secure package installation.
  • [EXTERNAL_DOWNLOADS]: References the 'secsi/getjs' Docker image and the official 'projectdiscovery/katana' Docker image for link analysis.
  • [COMMAND_EXECUTION]: Executes the 'katana' binary and 'docker' commands to perform active web crawling and JavaScript analysis.
  • [PRIVILEGE_ESCALATION]: Includes installation instructions for Ubuntu that utilize 'sudo' for updating package lists, installing prerequisites, and configuring system repositories.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: Processes content from external target websites through the 'katana' crawler (SKILL.md).
  • Capability inventory: Invokes shell commands ('katana', 'docker') and performs file writes to the 'artifacts/' directory (SKILL.md).
  • Boundary markers: No explicit instructions or delimiters are provided to the agent to prevent the interpretation of instructions found within crawled web data.
  • Sanitization: No sanitization or filtering of retrieved web content is documented before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 8, 2026, 07:58 AM