test
Pass
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted external data such as code diffs, pull requests, and URLs. An attacker could embed malicious instructions within these materials to manipulate the agent's actions.
- Ingestion points: The workflow instructions specify reading tasks, source designs, changed code, and external URLs (Workflow Step 1).
- Boundary markers: There are no defined delimiters or instructions to treat the external content as non-executable data.
- Capability inventory: The skill requires the ability to execute shell commands for "automated checks" and manage application processes.
- Sanitization: No methods for sanitizing or escaping external content are implemented before processing or interpolation into prompts.
- [COMMAND_EXECUTION]: The workflow requires the agent to execute automated checks and start the application under test (Workflow Steps 6 and 7). This capability is inherent to the skill's purpose but represents a potential exploitation path if the agent is compromised by malicious input from the code being tested.
Audit Metadata