office-document-specialist-suite-hardened

Pass

Audited by Gen Agent Trust Hub on Apr 21, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains adversarial prompt examples in SAFETY.md (e.g., 'ignore previous instructions') which are used for training and demonstrating defensive guardrails rather than for malicious intent. It also addresses the surface for indirect prompt injection by mandating explicit file specification and absolute path resolution in SKILL.md. Ingestion points: Document parsing in ods.py. Boundary markers: Explicit instructions in SKILL.md. Capability inventory: File system writes in ods.py, dependency installation in setup.sh. Sanitization: Content validation is deferred to the agent through guardrails.
  • [COMMAND_EXECUTION]: The setup.sh script executes shell commands to initialize a Python virtual environment and install declared dependencies from PyPI, following standard development practices.
  • [DATA_EXFILTRATION]: The skill identifies sensitive paths such as ~/.ssh/ and .env in its documentation to provide the agent with negative constraints, explicitly prohibiting unauthorized access to credential files.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 21, 2026, 06:31 PM
Security Audit — agent-trust-hub — office-document-specialist-suite-hardened