ai_llm_engineer
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- [NO_CODE] (SAFE): The provided file contains only Markdown text and YAML metadata. There are no scripts, binaries, or configuration files that execute logic on the host system.
- [PROMPT_INJECTION] (SAFE): The skill does not attempt to override system instructions. Conversely, it explicitly instructs the agent to maintain 'Prompt Injection protection' as a core rule.
- [DATA_EXPOSURE] (SAFE): No credentials, API keys, or sensitive file paths (e.g., .ssh, .env) are present in the documentation or metadata.
- [INDIRECT_PROMPT_INJECTION] (LOW): The skill's primary function is to process user requests to generate structured prompts. While it ingests untrusted data (user queries), it lacks the 'Capability' tier (no shell access, no network, no file writes) required for a High or Medium severity classification. The risk is limited to the integrity of the generated text output.
Audit Metadata