airtable-automation

Warn

Audited by Gen Agent Trust Hub on Aug 8, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires adding 'https://rube.app/mcp' as a remote MCP server. This connects the agent to a third-party service provider not recognized as a trusted vendor.
  • [DATA_EXFILTRATION]: The skill's architecture involves routing sensitive Airtable data, including records, comments, and base schemas, through the external domain 'rube.app'.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing data from Airtable (Ingestion points: AIRTABLE_LIST_RECORDS, AIRTABLE_GET_RECORD, AIRTABLE_LIST_COMMENTS in SKILL.md) and possessing write capabilities (Capabilities: AIRTABLE_CREATE_RECORD, AIRTABLE_UPDATE_RECORD, AIRTABLE_CREATE_FIELD in SKILL.md). The absence of boundary markers or sanitization logic allows malicious instructions within database records to potentially influence agent behavior.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 8, 2026, 12:54 PM
Security Audit — agent-trust-hub — airtable-automation