tighten-instruction
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes instruction lines from users or external files (such as
CLAUDE.md) and includes instructions to apply edits based on a confidence score. This creates an attack surface for indirect prompt injection if the processed content contains malicious instructions designed to influence the agent's behavior during the rewriting process. - Ingestion points: User input or content from files like
CLAUDE.mdand rule files. - Boundary markers: The skill does not define specific delimiters for the input text.
- Capability inventory: The skill suggests applying edits to files (file system write capability for the agent using this skill).
- Sanitization: No explicit validation or sanitization of the input text is mentioned.
Audit Metadata