deslop
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No patterns of prompt injection, role-play bypass, or safety filter overrides were detected in the instructions.
- [DATA_EXPOSURE_AND_EXFILTRATION]: No sensitive data, hardcoded credentials, or network exfiltration patterns were found.
- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill does not perform any remote code execution or package installations. It relies purely on Markdown reference files.
- [EXTERNAL_DOWNLOADS]: The README provides links to download releases from the author's GitHub repository. These are standard installation links for the skill itself.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (prose) to perform stylistic cleanup. 1. Ingestion points: User-provided text processed via triggers in SKILL.md. 2. Boundary markers: None present; the agent processes input directly based on provided rules. 3. Capability inventory: Limited to text rewriting and transformation; no shell access, file system writes, or network operations are involved. 4. Sanitization: No input validation or sanitization of user-provided prose is performed.
- [DYNAMIC_CONTEXT_INJECTION]: No use of shell command substitution syntax (!command) was found in the SKILL.md file.
Audit Metadata