kb-retriever
Pass
Audited by Gen Agent Trust Hub on May 4, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to utilize standard local command-line tools like
grep,pdftotext, andpdftoppm. These utilities are used as intended for searching text and converting documents within the user's workspace, with no evidence of command injection or malicious shell operations. - [EXTERNAL_DOWNLOADS]: The documentation mentions several common Python libraries for document processing, including
pandas,pdfplumber,pypdf, andpdf2image. These are reputable, standard packages from official registries and are necessary for the skill's primary function. - [PROMPT_INJECTION]: Since the skill retrieves and processes information from external documents to answer user queries, it has a surface for indirect prompt injection. However, the skill implements defensive practices by instructing the agent to read files in small, windowed chunks (using
limitandoffset) and providing specific reference guides to ensure tools are used correctly, which helps the agent treat document content as data context rather than instructions. - [DATA_EXFILTRATION]: No network operations or exfiltration patterns were detected. The skill operates strictly on local file paths and emphasizes processing data within the local environment.
Audit Metadata