cookie-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides command templates that interpolate untrusted data (e.g.,
<cookie_value>,<cookie>) directly into shell commands. This creates a vulnerability surface where a malicious input containing shell metacharacters could achieve arbitrary command execution on the agent's host. - Ingestion points: User-provided or target-extracted cookie values in
SKILL.md. - Boundary markers: Absent; commands use simple string interpolation.
- Capability inventory: Employs
curl,base64,python3, andflask-unsignfor data processing and network requests. - Sanitization: No input validation or escaping is suggested for the cookie values before they are passed to shell commands.
- [COMMAND_EXECUTION]: The methodology relies extensively on executing shell commands and Python scripts to perform security analysis and authentication bypass.
Audit Metadata