enterprise-itil-service-mgmt
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits extreme text bloating across eight reference files (e.g.,
architecture-patterns.md,deployment-pipelines.md,performance-optimization.md). Each file contains 150 sections of technical specifications that are largely identical or composed of redundant boilerplate text and equations. - The files contain hundreds of repetitions of the same code snippets, such as a Python
concurrent_fetchfunction and various Go/Rust templates. - This excessive repetition (totaling approximately 1,200 repetitive sections) acts as a context-filling technique that can exhaust the AI agent's token window, potentially displacing core instructions or confusing the model's logic. While no direct malicious payload was found, such bloating is an adversarial pattern used to bypass safety filters or degrade system performance.
Audit Metadata