Commute Matrix

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes external data from camp provider files and family profiles, which presents a surface for indirect prompt injection attacks.
  • Ingestion points: Reads all markdown files in the research providers directory and the family-profile.md file.
  • Boundary markers: Absent. The skill does not use specific delimiters or instructions to prevent the agent from obeying instructions embedded in the provider files.
  • Capability inventory: Executes a local Python script that performs network requests to an external API and writes updates to the local filesystem.
  • Sanitization: Uses yaml.safe_load for configuration parsing and URL encoding for API parameters; however, camp names and addresses are used directly in commands and output reports without validation.
  • [COMMAND_EXECUTION]: Runs a local Python script (commute_calculator.py) to process commute data. The script is invoked using system shell commands with arguments for file paths and API credentials.
  • [DATA_EXFILTRATION]: Transmits location data to the official Geoapify API (api.geoapify.com) to retrieve geocoding and routing information. This communication is a legitimate part of the skill's intended functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 11:31 AM
Security Audit — agent-trust-hub — Commute Matrix