What problem does it solve? Code quality issues are often caught too late—at commit time or in CI—forcing costly rework. This Skill enforces formatting and linting on every file edit through Claude Code hooks, automatically fixing violations before they accumulate and preventing agents from gaming linter configs. ## Core Features & Use Cases - Three-Phase Enforcement: Auto-formats silently, collects remaining violations as structured JSON, then delegates fixes to Claude subprocesses routed by complexity (Haiku for style, Sonnet for logic, Opus for type errors). - Config Protection: Blocks edits to linter configs like .ruff.toml and biome.json via PreToolUse hooks, and detects tampering at session end with a Stop hook. - Multi-Language Support: Covers Python, TypeScript, Shell, YAML, JSON, TOML, Markdown, and Dockerfile with tools like ruff, biome, shellcheck, and hadolint. - Use Case: While an agent edits a Python file, the hook runs ruff format, collects unresolved violations, spawns a Sonnet subprocess to fix a complexity warning, and only reports back if violations remain. ## Quick Start Set up Plankton hooks in my project so every file edit is automatically formatted and linted with violations fixed by model-tiered subprocesses.
