ripgrep-over-grep-command
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill uses authoritative directives such as "MANDATORY" and "NEVER" to instruct the agent to override its default behavior (shell-based searching) in favor of a specific protocol. While standard for technical documentation, these are categorized as instructional overrides.
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a protocol for code searching which inherently creates a surface for indirect prompt injection if the files being searched contain malicious instructions.
- Ingestion points: The agent processes file content retrieved through the native
grep_searchtool based on theQueryandSearchPathspecified in SKILL.md. - Boundary markers: The protocol uses a structured JSON schema for search results but does not specify delimiters or warnings to treat search hits as untrusted data.
- Capability inventory: The skill assumes access to a native
grep_searchtool andrun_commandcapability across several AI platforms. - Sanitization: There are no sanitization or escaping mechanisms defined for the search results or queries within the instruction set.
- [NO_CODE]: The skill consists entirely of instructional markdown and documentation; no executable scripts, shell commands, or configuration files are provided, which limits the risk of direct malicious execution.
Audit Metadata