meeting-automation

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill connects to the official Feishu (Lark) API at open.feishu.cn to retrieve meeting and room data. Communication is secured via HTTPS and targets a well-known enterprise service.\n- [CREDENTIALS_UNSAFE]: The skill appropriately manages the LARK_TENANT_TOKEN by retrieving it from environment variables, ensuring that no sensitive credentials are hardcoded in the source code.\n- [COMMAND_EXECUTION]: All logic is implemented using Python's standard libraries (urllib.request, json). There are no calls to system shells, subprocess, or other command execution utilities.\n- [PROMPT_INJECTION]: The skill processes external meeting metadata (titles, descriptions), which constitutes an indirect prompt injection surface. The risk is negligible as the skill is strictly read-only and lacks invasive system permissions.\n
  • Ingestion points: Meeting and room data retrieved from Feishu in run.py.\n
  • Boundary markers: None; data is passed directly into the agent context.\n
  • Capability inventory: No write access, file-system access, or code execution capabilities are present.\n
  • Sanitization: Relies on structured JSON responses without explicit per-field sanitization.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 12:36 AM