agent-qa-testing
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a shell script (
scripts/agent-qa.sh) that dynamically constructs and executes command-line instructions to prompt AI agents using theclaudeCLI. It reads agent system instructions from the local filesystem (agents/directory) to facilitate testing. - [INDIRECT_PROMPT_INJECTION]: As a testing framework, the skill ingests user-provided text inputs from YAML test suites and passes them to LLMs. This establishes a vulnerability surface where a malicious test case could attempt to inject instructions into the agent being evaluated, though this is an inherent risk in LLM testing utilities.
Audit Metadata