groq-security-basics
Warn
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: MEDIUMPERSISTENCEPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PERSISTENCE]: The skill provides instructions and code in references/examples.md to create a git pre-commit hook in .git/hooks/pre-commit. While intended for security scanning, this establishes a mechanism where code is executed automatically across development sessions.
- [PRIVILEGE_ESCALATION]: The implementation examples in references/examples.md include the use of chmod +x on the dynamically created pre-commit hook file. This increases the execution permissions of a file generated by the skill's instructions.
- [INDIRECT_PROMPT_INJECTION]: The skill documents a server-side proxy pattern in references/implementation.md that ingests untrusted user input from API requests. 1. Ingestion points: Next.js route handler. 2. Boundary markers: Suggested use of a HARDENED_SYSTEM_PROMPT to delimit instructions. 3. Capability inventory: The proxied request calls the Groq inference API. 4. Sanitization: A sanitizeUserInput function is provided to filter common injection keywords.
- [COMMAND_EXECUTION]: The skill provides several bash command sequences in references/examples.md for environment configuration, secret management using official CLI tools (AWS, GCP, Vercel, GitHub), and filesystem auditing using grep. These commands are legitimate for the skill's stated purpose but involve direct shell interaction.
- [PROMPT_INJECTION]: Documentation in SKILL.md and references/implementation.md contains strings typically associated with prompt injection (e.g., 'ignore previous instructions'). Analysis confirms these are defensive examples and sanitization logic provided to the user, not active injection attempts.
Audit Metadata