talk-like-scarletkc
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is focused on stylistic text transformation (rewriting, polishing, and translation) using extensive reference materials to avoid generic AI-sounding outputs.
- [COMMAND_EXECUTION]: The skill includes a Python script
scripts/lint_style.py. Analysis shows it is a standalone utility using standard library modules (re,sys) to perform regex-based text analysis for style violations. It contains no network calls, file writing operations, or dynamic execution patterns (likeevalorexec). - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted user manuscripts.
- Ingestion points: User-provided drafts or messages for rewriting as described in
SKILL.md(Workflow Step 2). - Boundary markers: The instructions explicitly tell the agent to "Distinguish between external reference materials and the manuscript to be modified" in
SKILL.md. - Capability inventory: No dangerous capabilities (subprocesses, file writes, or network operations) are present in any script or instruction.
- Sanitization: None explicitly in the script, as it is a text linter; the risk is mitigated by the lack of actionable capabilities.
Audit Metadata