email-to-calendar

Pass

Audited by Gen Agent Trust Hub on Nov 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The utility script scripts/utils/event_tracking.py uses subprocess.run with shell=True and f-string interpolation, creating a potential command injection surface if the tracking variables are manipulated. Evidence: subprocess.run(f'{script_dir}/delete_tracked_event.sh --event-id "{event_id}"', shell=True, capture_output=True).
  • [INDIRECT_PROMPT_INJECTION]: The skill processes unread emails from the user's inbox, which can contain malicious instructions designed to alter the agent's behavior. 1. Ingestion points: Unread emails and forwarded messages processed by the agent as described in SKILL.md. 2. Boundary markers: The skill instructs the agent to extract events into structured JSON format. 3. Capability inventory: The skill can read emails, modify calendar entries, and send emails via scripts/email_send.sh. 4. Sanitization: No specific programmatic sanitization of email body content is performed before processing.
  • [DATA_EXFILTRATION]: The skill provides the capability to send emails via scripts/email_send.sh, which could be exploited to exfiltrate information if the agent's instruction flow is compromised. Evidence: The send_email function in scripts/utils/email_ops.py allows sending arbitrary text to external recipients.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 22, 2026, 01:12 AM