add-reference

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection from external sources.
  • Ingestion points: In Step 3, the skill fetches content from user-provided URLs and reads user-supplied files.
  • Boundary markers: There are no instructions to use delimiters (like XML tags or triple quotes) or 'ignore instructions' markers when summarizing or processing external content.
  • Capability inventory: The skill has access to Bash, Write, and Read tools, which could be exploited if an agent accidentally obeys instructions found within the fetched content.
  • Sanitization: No sanitization, escaping, or validation of the fetched URL content or file metadata is performed before writing it to the project's reference directory.
  • [COMMAND_EXECUTION]: The skill executes shell commands using the Bash tool.
  • Step 1 uses 'mkdir -p' to ensure the existence of the project's reference directory.
  • [EXTERNAL_DOWNLOADS]: The skill performs network operations to retrieve external data.
  • Step 3 includes a process to 'Fetch the URL content' for any URL provided by the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 07:13 PM