esp32-debugging
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it processes untrusted user or application error logs and has access to active capabilities.
- Ingestion points: Untrusted external data enters the agent context through the
$ARGUMENTSvariable or user-supplied serial monitor and panic logs inSKILL.md. - Boundary markers: There are no explicit boundaries or instructions defined to isolate or prevent the interpretation of malicious instructions embedded within the error text.
- Capability inventory: The skill possesses shell execution capabilities using the
Bash(just:*)tool to run project compilation and monitoring tasks inSKILL.md. - Sanitization: No sanitization, escaping, or strict validation mechanisms are applied to the input before it is analyzed by the agent.
Audit Metadata