healing-word

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious code or patterns detected. The skill instructions are benign and focus on technical remediation workflows.
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by processing external issue descriptions to generate technical commands.
  • Ingestion points: User-provided issue, error, or problem descriptions (SKILL.md).
  • Boundary markers: Absent; the prompt template does not define specific delimiters for untrusted data.
  • Capability inventory: The skill is designed to formulate executable instructions, config changes, or commands.
  • Sanitization: No explicit sanitization or validation of input data is defined in the procedure.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 07:47 PM
Security Audit — agent-trust-hub — healing-word