AI Context Linter

What it does

AI Context Linter runs in your GitHub workflow and looks for the files that tell AI coding agents how to behave in your repository, including CLAUDE.md, .cursorrules, AGENTS.md, COPILOT.md, .windsurfrules and .clinerules. It finds them automatically once the action is added to a workflow. Its checks fall into groups. Security findings, such as hardcoded API keys or leaked home-directory paths, are errors that fail the build. Structure findings cover files that are too long, lack headings or have empty sections. Anti-pattern checks flag vague, contradictory or redundant instructions, and lower-priority suggestions cover things like missing style guidance or large hidden HTML comments. Inputs let you choose which files to check, set a token limit and decide whether warnings fail the build, and an optional YAML config file can disable individual rules. The action reports how many issues were found and how many files were checked.

Capabilities

• Auto-detects common AI context files
• Catches hardcoded API keys and leaked private paths
• Flags oversized, unstructured or empty-section files
• Flags vague, conflicting and redundant instructions
• Configurable token limit and disabled rules
• Outputs issue and file counts for later workflow steps

When to use it

• You maintain CLAUDE.md or .cursorrules files and want them checked on each push or pull request
• You want to stop secrets from leaking into agent instruction files
• You want a quick quality check on context files shared across a team repository