config-lint

You can validate AI assistant configuration files by running a unified multi-linter pass that checks Claude Code skills, hooks, MCP servers, GitHub Copilot instructions, and CLAUDE.md files against official Anthropic and GitHub standards, generating a report with severity tags and remediation guidance.

Yes, you can automatically fix structural issues in Claude Code hook configurations and MCP settings by running the linter with the auto-remediation flag, which corrects problems like deprecated hook output formats and missing required frontmatter fields.

The best way to audit GitHub Copilot instruction files before committing is to use a unified config linter that checks for governance standards, hardcoded secrets, and structural compliance across both local and team-shared project environments in a single pass.

Yes, config linting supports read-only validation and targeted linting of specific AI assistant config types, allowing you to audit Claude Code skill definitions, hook configurations, MCP server settings, or CLAUDE.md project memory files individually without applying automatic fixes.

You should lint CLAUDE.md project memory files for team-shared projects because manual validation is error-prone, and linting ensures configuration consistency, prevents common pitfalls like invalid hook matchers, and enforces compliance with official best practices across the team.