The Agent Skills Directory

[SAFE]: The skill acts as a passive audit and education resource for Flutter development. Analysis of all skill files confirms that the behavior is limited to project-local code scanning and instructional guidance.\n- [COMMAND_EXECUTION]: The skill includes a local utility script, scripts/audit.sh, which assists users in scanning their Flutter projects. The script utilizes standard command-line tools such as grep and wc to analyze file contents for specific design patterns. It does not require root privileges, perform network operations, or access sensitive system files.\n- [PROMPT_INJECTION]: Indirect Prompt Injection analysis: 1. Ingestion points: Codebase files scanned via audit.sh and code snippets provided by users. 2. Boundary markers: No specific delimiters or markers are used to encapsulate ingested code. 3. Capability inventory: Read-only file system access through grep and shell utilities. 4. Sanitization: No content-based sanitization is performed on ingested code. This surface is characteristic of audit tools and is considered safe in this context.

flutter-taste