prompt-optimizer
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from user-provided draft prompts and project configuration files (such as
CLAUDE.md,package.json, orgo.mod) to generate optimized instructions for the agent. - Ingestion points: Phase 0 reads the content of
CLAUDE.mdand checks for the presence of various project configuration files. The user's input prompt is also a primary ingestion point. - Boundary markers: The instructions do not define explicit boundary markers or delimiters to separate untrusted external content from the internal optimization logic.
- Capability inventory: The skill is limited to reading file system metadata and specific project files. It contains explicit instructions prohibiting the agent from writing code, creating files, or executing implementational commands.
- Sanitization: There is no evidence of input sanitization, validation, or filtering of the ingested content from the user or the local environment.
Audit Metadata