things-mac

Warn

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads source code from a third-party, non-trusted GitHub repository (github.com/ossianhempel/things3-cli) during the installation phase.
  • [REMOTE_CODE_EXECUTION]: The installation process uses go install github.com/ossianhempel/things3-cli/cmd/things@latest, which fetches, compiles, and installs code from a remote source not included in the trusted vendors list.
  • [COMMAND_EXECUTION]: The skill relies on executing the things binary to perform its primary functions, including reading local database files and sending data via macOS URL schemes.
  • [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection.
  • Ingestion points: Data is read from the local Things 3 database via commands like things inbox, things today, and things search (SKILL.md).
  • Boundary markers: None identified; database content (titles, notes) is likely interpolated directly into the agent's context.
  • Capability inventory: The skill can execute CLI commands and potentially send network requests via the URL scheme handler.
  • Sanitization: There is no evidence of sanitization or escaping of the task titles or notes retrieved from the database.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 23, 2026, 12:56 PM