campaign-orchestrator

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • CREDENTIALS_UNSAFE (MEDIUM): The skill documentation specifies requiring a plaintext password (GOG_KEYRING_PASSWORD) in environment variables for Gmail access, which is an unsafe practice susceptible to exposure.
  • DATA_EXFILTRATION (MEDIUM): The webhook_handler.py script starts an unauthenticated network service using HTTPServer to receive external webhook data, creating a potential attack vector for unauthorized state manipulation.
  • COMMAND_EXECUTION (MEDIUM): The skill uses the system cron service for task persistence and schedules the execution of external CLI tools and Python scripts (e.g., gog-shapescale, attio) that are not provided in the source files, preventing a full audit of executed commands.
  • PROMPT_INJECTION (LOW): Indirect Prompt Injection surface detected. 1. Ingestion points: webhook_handler.py POST body containing SMS text. 2. Boundary markers: Absent; data is stored directly in campaigns.json. 3. Capability inventory: Automated SMS, email, and CRM command execution. 4. Sanitization: Absent; the raw message content is processed and saved without filtering.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 08:07 PM