macos-terminal
Warn
Audited by Gen Agent Trust Hub on Mar 22, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/window/run-script.applescriptuses the AppleScriptdo scriptcommand to execute arbitrary text as shell commands in Terminal.app. This allows an agent to perform any action on the system that the current user has permissions for. - [DATA_EXFILTRATION]: Scripts such as
scripts/tab/contents.applescriptandscripts/tab/history.applescriptallow an agent to retrieve the entire scrollback buffer and session history of terminal tabs. This can expose sensitive data like API keys, environment variables, or private file contents previously displayed in the terminal. - [PROMPT_INJECTION]: The skill facilitates indirect prompt injection by reading raw terminal output which could contain malicious instructions from external sources.
- Ingestion points:
scripts/tab/contents.applescript,scripts/tab/history.applescript. - Boundary markers: None present in the data retrieval scripts.
- Capability inventory:
scripts/window/run-script.applescript(arbitrary command execution). - Sanitization: None; raw terminal contents are passed directly to the agent context.
Audit Metadata