github-url-intercept

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill uses strong imperative language such as 'BLOCKING INTERCEPT' and 'STOP' to override the agent's default behavior for handling web links. While functional for token optimization, these markers match instruction-overriding patterns.
  • [PROMPT_INJECTION]: The skill processes untrusted GitHub URLs, creating an indirect prompt injection surface. Ingestion points: External URLs from user prompts are parsed by the scripts/test_url_routing.py utility. Boundary markers: The skill instructs the agent to return and process structured JSON output. Capability inventory: The skill can execute local Python scripts and the GitHub CLI (gh api). Sanitization: The routing script implements specific regex validation and character filtering (CWE-78 mitigation) to prevent command injection from malformed URL paths or repository names.
  • [COMMAND_EXECUTION]: The skill provides the agent with specific shell commands to execute python3 scripts and gh api requests using parameters extracted from user-provided URLs.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 02:42 AM