semgrep-rule-variant-creator
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the
semgrepCLI for various tasks including testing (--test), validation (--validate), and AST analysis (--dump-ast). It also uses standard filesystem commands likelsandfindto locate reference materials within the agent's environment.- [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for indirect prompt injection when processing external Semgrep rule files and source code. - Ingestion points: Files located at the provided
rulePathand source code analyzed during the validation phase. - Boundary markers: The skill implements a structured four-phase workflow (Applicability, Testing, Translation, Validation) to partition tasks and limit the scope of agent actions.
- Capability inventory: Uses
semgrepfor structural analysis and testing, alongside standard file read/write operations. - Sanitization: Employs structural parsing via the Semgrep engine to evaluate rule patterns rather than treating the input as executable logic.- [DYNAMIC_EXECUTION]: The
language-syntax-guide.mddocumentation contains references to dynamic execution patterns such aseval($USER_INPUT). These are provided strictly as examples for writing Semgrep rules to detect vulnerabilities in other codebases and do not constitute executable code within the skill context.
Audit Metadata