moai-security-compliance

Warn

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references several third-party Node.js packages and external APIs to facilitate compliance monitoring.
  • It utilizes the drata-api for integration with Drata, a compliance automation service.
  • It utilizes context7-mcp, which is an external dependency for regulatory framework checks.
  • [COMMAND_EXECUTION]: The skill's configuration in SKILL.md explicitly enables the Bash tool. While no bash scripts are provided in the analyzed files, this permission provides the agent with a powerful capability to execute shell commands on the host environment.
  • [PROMPT_INJECTION]: The skill processes external data for classification (PII, PHI, payment data) and logging, creating a surface for indirect prompt injection.
  • Ingestion points: DataClassifier.classify() and AuditLogger.logDataAccess() ingest potentially untrusted data fields for analysis and logging in SKILL.md and examples.md.
  • Boundary markers: The processing logic does not implement boundary markers or instructions to disregard embedded commands in the data streams.
  • Capability inventory: The skill possesses capabilities for file system reads (fs.readFile), database deletions (db.users.deleteOne), and network operations via AWS and Drata APIs.
  • Sanitization: Data classification relies on regex patterns (/email|phone|address/) and uses JSON.stringify(), but lacks sanitization to prevent the agent from interpreting instructions contained within the data values.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 1, 2026, 01:07 AM