The Agent Skills Directory

[DATA_EXPOSURE]: The skill references sensitive file paths such as ~/.ssh/id_* and .env. However, these are mentioned exclusively in the context of standard troubleshooting and security maintenance (e.g., setting correct 600 permissions for SSH keys or creating a .env file from a template). This follows industry best practices for local environment management.
[COMMAND_EXECUTION]: The skill suggests several common diagnostic commands (e.g., lsof, kill, npm, pip, docker). These are standard utility commands used for development environment debugging and do not perform unexpected or hidden actions.
[PRIVILEGE_ESCALATION]: The skill explicitly advises against the unsafe use of elevated privileges (e.g., recommending against using sudo for npm install), which enhances the security posture of the user's environment.
[INDIRECT_PROMPT_INJECTION]: As a troubleshooting tool, the skill is designed to process external data such as error logs and terminal output. While this represents a theoretical ingestion surface for indirect instructions, the context is limited to diagnostic commands and no automated exploitation patterns were identified.
Ingestion points: Terminal error messages and service logs.
Boundary markers: None explicitly defined in the instructions.
Capability inventory: File system permissions check (ls, chmod), process management (lsof, kill), and package management (npm, pip).
Sanitization: None specified, relying on standard agent execution guardrails.

env-troubleshoot