ace-search
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION] (LOW): The skill instructs the agent to use
grepandrgfor text-based searching within the local codebase. While standard for developer tools, this involves executing shell commands. - [PROMPT_INJECTION] (LOW): The skill possesses an indirect prompt injection surface (Category 8) because it retrieves and processes content from an external codebase. Evidence Chain: 1. Ingestion points: Codebase content via ACE retrieval and shell search tools in file SKILL.md. 2. Boundary markers: Absent; instructions do not define delimiters to isolate code content from instructions. 3. Capability inventory: Limited to read-only search and retrieval. 4. Sanitization: Absent; results are summarized from raw file content. Tier: INFO (External content + display only).
Audit Metadata