llm-auditor
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious code, obfuscation, or unauthorized data access patterns were detected. All components align with the skill's stated purpose of auditing factual claims.- [PROMPT_INJECTION]: The skill is designed to process and analyze external text or previous model outputs, which is a characteristic surface for indirect prompt injection.
- Ingestion points: Processes provided text or previous responses for claim extraction in Phase 1 (SKILL.md).
- Boundary markers: No specific delimiters are defined to isolate the audited text from the instructions.
- Capability inventory: Utilizes the web_search tool to verify claims (SKILL.md).
- Sanitization: No specific sanitization or filtering of the input text is implemented.- [EXTERNAL_DOWNLOADS]: The skill includes an attribution link to an official Google repository for its architectural design.
- Source: github.com/google/adk-samples
Audit Metadata