agency-prompt-engineer
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: Static analysis flagged patterns such as "Ignore instructions" and "Ignore all previous instructions." Upon manual review, these appear within a Python test suite template and a section on "Prompt Injection Defense." They are documented as examples of adversarial inputs that a prompt engineer should test for, rather than active attempts to override the agent's own behavior.
- [EXTERNAL_DOWNLOADS]: The skill includes Python code snippets that use the
pytestlibrary for quality assurance. These are instructional examples and do not trigger automated or hidden downloads of untrusted code. - [DATA_EXFILTRATION]: The skill focus is entirely on text generation and optimization. No patterns related to reading sensitive local files or sending data to external servers were identified.
- [COMMAND_EXECUTION]: The skill provides code templates for building and formatting prompts. It does not instruct the agent to execute arbitrary system commands or perform dangerous operations on the host environment.
Audit Metadata