rulecheck

Automated rule compliance is enforced by scanning the codebase for violations, fixing the highest-impact ones in an isolated worktree, running validation, and creating a pull request. This uses a memory-enabled agent to ensure auditable and repeatable outcomes.

Yes, you can fix code violations in an isolated git worktree. The workflow detects violations, applies fixes safely within the worktree to avoid affecting the main branch, runs full validation, and creates a PR.

Autonomous code-rule checking with persistent memory uses a memory-enabled agent to scan for rule violations, apply fixes, and track progress across multiple runs. This persistent memory improves future runs by remembering past violations and applied solutions.

Yes, autonomous rule checking works across repositories of varying sizes. The deterministic workflow operates effectively on different repository scales, enabling teams to enforce coding standards consistently regardless of project size.

To start an autonomous code review run, invoke the skill with an optional focus area. The agent will then scan the codebase for CLAUDE.md violations, fix them in an isolated worktree, validate the changes, and create a PR.

The deterministic workflow for fixing rule violations is limited by the rules defined in CLAUDE.md and the agent's interpretation of them. Complex, ambiguous rule violations may require manual review, and the automated fixes are constrained by the applied deterministic logic.