frozen-checks
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill defines a normative grammar for executing shell commands via a runner (e.g.,
- RUN:command-> exit:<n>). It explicitly instructs the agent to run these drafted commands against the codebase tree to verify they function as expected before 'freezing' the check. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of issue descriptions and decomposition data to generate these check commands. This creates an attack surface where malicious input in an issue or codebase could potentially lead to the generation of unintended shell commands.
- Ingestion points: Data used during issue decomposition and check drafting.
- Boundary markers: Commands are encapsulated in backticks within a specific
RUNsyntax, which provides structural delimitation. - Capability inventory: Execution of shell commands via the
check-runner.ps1script. - Sanitization: No explicit sanitization or filtering of external content is mentioned beyond the structural grammar.
Audit Metadata