ast-grep
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process source code from the repository using structural search patterns, which presents a risk of indirect prompt injection if the code contains malicious instructions or structural patterns intended to mislead the agent.
- Ingestion points: Source code files processed by
ast-grepsearch and rewrite commands. - Boundary markers: The workflow relies on the agent's interpretation of tool output and specifies a manual diff review, but lacks strict boundary markers or sanitization for code content displayed in the context.
- Capability inventory: The skill possesses the capability to perform repository-wide code searches and modifications (rewrites) through the
ast-greputility. - Sanitization: The instructions in
references/safety-and-review.mdmandate a "preview-before-apply" workflow and a review checklist to mitigate unintended changes. - [DYNAMIC_EXECUTION]: The skill includes a helper script
scripts/scaffold-ast-grep-command.mjsthat dynamically constructs shell commands based on user-provided arguments. - Evidence: The Node.js script uses string concatenation to assemble
ast-grepcommand lines from input variables such as patterns, rewrites, and globs. While it utilizes ashellQuotefunction for escaping, the runtime generation of executable command strings is a characteristic of dynamic execution.
Audit Metadata