llm-debugger
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- PROMPT_INJECTION (LOW): The skill is susceptible to indirect prompt injection (Category 8) because it ingests untrusted data and interpolates it into new prompts without sanitization.
- Ingestion points: Functions like
diagnose_failure,generate_test_cases, andsuggest_prompt_fixaccept raw string inputs from potentially untrusted LLM outputs. - Boundary markers: There are no delimiters (like XML tags or markdown blocks) or specific 'ignore embedded instructions' warnings used when processing the data.
- Capability inventory: The skill code consists entirely of logic and string manipulation. No dangerous capabilities such as
os.system,subprocess, file writing, or network requests are present. - Sanitization: The skill does not perform any escaping or validation on the
inputoroutputstrings before concatenating them into the final prompt or test case objects.
Audit Metadata