What problem does it solve? Teams adopt lint tooling but never decide which rules to enable, leaving coding standards as unstated preferences. This Skill installs a versioned, pre-rendered rule selection for Python (ruff) or React (ESLint) into the configuration the repository's linter already reads, with provenance stamps and an explicit record of what was chosen. ## Core Features & Use Cases - Stack inference from files: Detects applicable profiles from pyproject.toml, package.json, and tsconfig.json rather than self-declarations, and reports exactly which predicate failed when nothing applies. - Consent-based profile selection: Presents rule counts, fix availability, and unenforced residue per level from a shipped manifest, then requires an explicit yes before any write. - Safe, idempotent writes: Copies byte-identical rendered artefacts inside marked spans, refuses to loosen existing selections or overwrite keys owned by others, and pins the installed versions in .claude/skill-config.yaml so re-runs change nothing. - Use Case: A repository with ruff wired in CI but no agreed rule set runs the skill, consents to python/standard, and gets a stamped [tool.ruff] region in pyproject.toml plus a written list of the properties no linter can enforce. ## Quick Start Ask the assistant to install a coding standard for this repository and confirm the profile level when prompted.