waitsec-core

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as user prompts and stack traces. It mitigates risks by acting as a constraint layer that requires the agent to verify assumptions and implement security best practices in generated code.
  • Ingestion points: User prompts and stack traces (SKILL.md, debug-first.md).
  • Boundary markers: The skill itself is a guardrail framework requiring clarification of ambiguous inputs.
  • Capability inventory: File editing and shell execution for testing/verification.
  • Sanitization: Mandates input validation and output escaping in anti-overengineering.md.
  • [COMMAND_EXECUTION]: The skill advocates for running terminal commands (e.g., npm test, curl, pytest) to verify code correctness and debug issues. These are standard developer tools used within a controlled verification sequence.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 08:36 AM
Security Audit — agent-trust-hub — waitsec-core