skills/alibaba/neug/generate-testcase/Gen Agent Trust Hub

generate-testcase

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from git commits and user arguments to generate and execute code.
  • Ingestion points: User input via $ARGUMENTS and file content/history from git commits (SKILL.md).
  • Boundary markers: No delimiters or isolation instructions are present to separate untrusted commit data from the agent's logic.
  • Capability inventory: The skill is designed to build and run test cases, which involves local command execution (SKILL.md).
  • Sanitization: The skill does not perform any validation or sanitization of the content extracted from commits before using it to generate executable test functions.
  • [COMMAND_EXECUTION]: The skill requires the agent to perform local command execution to build the project, run tests, and generate coverage reports using tools like fastcov and lcov (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 10:27 AM
Security Audit — agent-trust-hub — generate-testcase