ics-calendar-reader

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMPROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
  • PROMPT_INJECTION (MEDIUM): The skill is designed to ingest and parse untrusted data from external iCalendar (.ics) files, creating a surface for indirect prompt injection attacks. Malicious instructions embedded in calendar fields could override the agent's primary instructions.\n
  • Ingestion points: The scripts/read_ics.py tool fetches content from URLs defined in the ICS_URLS environment variable (file: SKILL.md).\n
  • Boundary markers: No boundary markers or delimiters are specified to separate untrusted event data from agent instructions. The skill explicitly instructs the agent to treat the parsed JSON as the 'source of truth'.\n
  • Capability inventory: The agent is instructed to summarize and filter the parsed event data. The underlying script has the capability to make network requests to arbitrary URLs provided in the environment.\n
  • Sanitization: No sanitization or filtering of the content within the SUMMARY, DESCRIPTION, or other ICS fields is mentioned or implemented in the provided configuration.\n- COMMAND_EXECUTION (LOW): The skill workflow involves executing a Python script (scripts/read_ics.py) with various command-line arguments to parse data. This is standard functionality but represents a local execution surface.\n- NO_CODE (LOW): The executable script scripts/read_ics.py referenced in the documentation is missing from the skill package, which prevents a full security audit of the implementation logic and its handling of malformed ICS data.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 09:56 AM