Chat with PDF

Pass

Audited by Gen Agent Trust Hub on Jun 3, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill instructions are focused exclusively on document interaction and do not attempt to override system prompts or safety guardrails.
  • [SAFE]: No sensitive file paths (e.g., .ssh, .env) or hardcoded credentials were found in the skill content.
  • [SAFE]: The skill uses a defined MCP server ('office-mcp') to perform its tasks, which is a standard method for extending agent functionality without resorting to arbitrary command execution.
  • [SAFE]: No obfuscation techniques, such as Base64 encoding of commands or hidden Unicode characters, were detected.
  • [PROMPT_INJECTION]: The skill processes untrusted PDF data, which is an inherent surface for indirect prompt injection.
  • Ingestion points: Document content is ingested via the 'extract_text_from_pdf' tool defined in SKILL.md.
  • Boundary markers: Absent; the skill does not instruct the agent to use specific delimiters to separate PDF content from instructions.
  • Capability inventory: The available tools are limited to text extraction ('extract_text_from_pdf') and metadata retrieval ('get_pdf_metadata'). There are no high-risk capabilities such as network access or shell execution.
  • Sanitization: No content validation or sanitization of the extracted text is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 3, 2026, 02:51 PM
Security Audit — agent-trust-hub — Chat with PDF