experiment
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
Bashtool to executeasta experiment, which runs software based on user-provided task parameters. - [DYNAMIC_EXECUTION]: The skill is designed to generate and execute software at runtime to answer research questions, which is the primary intended behavior.
- [INDIRECT_PROMPT_INJECTION]: 1. Ingestion points: Processes user-provided
TASKandBACKGROUND_KNOWLEDGEstrings, as well as contents fromTASK_FILEandBACKGROUND_KNOWLEDGE_FILE(SKILL.md). 2. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the command construction. 3. Capability inventory: The skill possesses shell execution capabilities throughBash(asta experiment *)andBash(mkdir -p ...)(SKILL.md), which are used to generate and execute software. 4. Sanitization: No sanitization of the user-provided strings or file content is performed before they are interpolated into the shell command execution.
Audit Metadata