The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes platform-native PreToolUse and PostToolUse hooks to execute shell scripts such as multi_linter.sh, protect_linter_configs.sh, and stop_config_guardian.sh. These scripts run automatically during file edits to ensure code quality and configuration integrity.
[REMOTE_CODE_EXECUTION]: The Phase 3 'Delegate + Verify' workflow involves spawning claude -p subprocesses. These subprocesses are used to route code violations to specific model tiers (Haiku, Sonnet, or Opus) based on complexity to perform automated repairs.
[EXTERNAL_DOWNLOADS]: The skill requires several external dependencies for its formatting and linting functions, including tools like ruff, uv, biome, shellcheck, and hadolint, typically installed via system package managers or language-specific registries.
[COMMAND_EXECUTION]: A specific PreToolUse hook is implemented to monitor and block legacy package manager commands (e.g., pip, npm, poetry) in favor of preferred alternatives like uv or bun, serving as a workflow enforcement mechanism.

plankton-code-quality