The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes environment detection commands such as where, command -v, and Get-Command to identify available shells and tools. These are standard system discovery operations and are constrained by a security boundary that explicitly prohibits executing business logic (linting, testing, deployments).
[DATA_EXPOSURE]: The skill creates and updates a local file (.terminal.local.md) to store system metadata. It includes explicit safeguards against reading or outputting sensitive data like keys, tokens, or private configurations.
[PRIVILEGE_ESCALATION]: No commands for privilege escalation (sudo, chmod 777) were detected. The skill operates within the user's current execution context.
[REMOTE_CODE_EXECUTION]: The skill does not download or execute remote scripts or external packages.
[PROMPT_INJECTION]: The instructions do not contain attempts to override system prompts or bypass safety guardrails.
[INDIRECT_PROMPT_INJECTION]: The skill reads from a local .terminal.local.md file if it exists, but the instructions require manual re-validation of any content read from the file before treating it as authoritative, mitigating the risk of obedience to malicious data injected into that file.

yy-detect-terminal