The Agent Skills Directory

[SAFE]: The skill implements a service management engine using tmux sessions. It does not perform unauthorized network requests or data exfiltration.- [COMMAND_EXECUTION]: The core engine executes user-defined project launcher scripts. It incorporates strong security controls: it uses physical path normalization (cd -P) to resolve paths, explicitly blocks execution via symbolic links, and employs shell-safe escaping (%q) to prevent command injection when passing parameters to tmux sessions.
[CREDENTIALS_UNSAFE]: The skill documentation correctly identifies the risk of hardcoded secrets and provides explicit instructions for users to manage credentials via project-level environment variables rather than the central configuration file.- [REMOTE_CODE_EXECUTION]: The installation and upgrade scripts operate locally and include syntax validation (bash -n) and configuration checks before applying changes. No remote script downloading or piped execution patterns were detected.- [SAFE]: The skill includes an attack surface for indirect instructions via its configuration file (~/.tmux.services). Security posture: [1] Ingestion: Loads service definitions from a shell-based config. [2] Boundaries: Uses explicit markers to isolate user-configured services from the engine logic. [3] Capabilities: Limited to tmux session management and local script execution. [4] Sanitization: Implements slug validation for service names and path constraints for launchers.

tmux-serv