targeted-line-range-edits
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill includes an example JSON snippet containing a hardcoded absolute Windows file path (c:/Users/ASUS/Documents/Newfolder/Antigravity/Major AI Skills/src/auth.ts), which exposes the local user account name and directory structure.
- [INDIRECT_PROMPT_INJECTION]: The skill's core purpose is instructing an agent on how to modify file contents, creating a surface for processing potentially malicious data. Ingestion points: Files processed by the replace_file_content tool. Boundary markers: The protocol uses StartLine and EndLine parameters to constrain the scope of edits. Capability inventory: The tool used is replace_file_content, which provides file system modification capabilities. Sanitization: There is no mention of sanitizing or validating the ReplacementContent before applying it to the file.
Audit Metadata