yzr-polish-writing
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill follows a structured workflow for text transformation using local reference files (
references/*.md) to guide the rewriting process. It does not perform network operations or execute arbitrary code. - [INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary text input provided by users or read from local files, which represents an indirect prompt injection surface.
- Ingestion points: Text snippets from chat history and local file content (via provided paths) in
SKILL.mdStep 1. - Boundary markers: None explicitly defined to delimit the content being polished from the agent's system instructions.
- Capability inventory: File system access for reading and in-place editing of local files.
- Sanitization: No explicit sanitization or filtering is performed on the input text before processing.
- [COMMAND_EXECUTION]: The skill describes a capability to perform "in-place editing" of local files. While this is a standard feature for a writing tool, it grants the agent the ability to modify the local filesystem based on user-specified paths.
Audit Metadata