Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    cursor/ CLAUDE.md Rules Conflicts & Overlap Auditor

    1

    Finds and fixes conflicting, redundant, and dead instructions in Cursor and Claude Code rule files.

    $12

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    cursor/ CLAUDE.md Rules Conflicts & Overlap Auditor

    cursor/ CLAUDE.md Rules Conflicts & Overlap Auditor

    Example session with this skill installed

    Audit my .cursorrules and CLAUDE.md. The agent keeps using npm even though CLAUDE.md says to use pnpm. Files are in the root.

    • Read your context and instructions
    • Compiled the cursor claude md

    I have audited .cursorrules and CLAUDE.md. I found a high-severity conflict regarding package management and two redundant style rules.

    Audit Report

    Inventory

    • .cursorrules: Global scope, 42 lines.
    • CLAUDE.md: Global scope, 115 lines.

    Confirmed Findings

    IDFindingSeverityEvidenceFix
    R001Direct ConflictHigh.cursorrules:12 ("Always use npm for installs") vs CLAUDE.md:28 ("Use pnpm for all package operations")Delete .cursorrules:12. CLAUDE.md is the repo's source of truth for scripts.
    R002RedundancyLow.cursorrules:5 ("Use ES6 modules") vs CLAUDE.md:14 ("Prefer ESM over CommonJS")Remove .cursorrules:5. Consolidate into CLAUDE.md coding standards.
    R003Dead RuleInfoCLAUDE.md:90 (Refers to scripts/test.sh which does not exist)Remove line 90 or update to npm test.

    Recommended Fix Plan

    1. Resolve Package Manager Conflict
    Delete the following from .cursorrules:

    - Always use npm for installs

    2. Consolidate Module Standards
    Update CLAUDE.md line 14:

    - Use ESM (ES6 modules) for all new files; avoid CommonJS.

    3. Cleanup
    Remove the dead reference to scripts/test.sh in CLAUDE.md.

    Next steps

    1. Apply the deletions to .cursorrules to stop the agent from defaulting to npm.
    2. Verify if any .cursor/rules/*.mdc files exist that might override these globals.
    3. Restart your agent session to reload the cleaned instructions.

    Disclaimer: This audit is advisory. Findings are based only on the files provided and tool behavior known at the time of audit. Re-verify before relying on precedence.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Rule files for AI coding agents often grow bloated, redundant, or contradictory as teams copy-paste instructions across different tools. This skill identifies precisely where your instructions clash or overlap, providing a prioritized plan to restore agent reliability and reduce token waste.

    What it does

    • Inventory rules by scanning CLAUDE.md, .cursorrules, .cursor/rules/*.mdc, AGENTS.md, and other instruction files to map their scope and line counts.
    • Detect conflicts such as direct contradictions, exclusive-choice clashes, and scope-shadowing where global rules fight specific globs.
    • Identify redundancies to find duplicates, near-duplicates, and subsumed rules that waste context window space.
    • Audit hygiene to flag dead rules matching no files, broken imports, and vague, unenforceable instructions.
    • Prioritize fixes based on severity, distinguishing between high-impact contradictions and low-priority style redundancies.

    How it works

    1. Discovery involves listing every rule file, its tool affinity, and its application scope.
    2. Deconstruction breaks files into atomic directives with unique IDs and normalized forms for comparison.
    3. Analysis applies a conflict taxonomy to find contradictions and overlaps, verified against the actual repository structure.
    4. Validation requires quoting exact file paths and line numbers for every finding to ensure accuracy.
    5. Reporting delivers a structured fix plan with ready-to-paste edits and a consolidated rule architecture.

    Frameworks & tools

    Works with Cursor, Claude Code, Windsurf, GitHub Copilot, and any agent using persistent instruction files like CLAUDE.md or .cursorrules.

    Why this beats prompting it yourself

    Generic prompts often miss subtle precedence logic or fail to cite specific line numbers, leading to "hallucinated" fixes. This skill uses a deterministic precedence model and strict evidence-based verification to ensure every recommendation is safe to apply.

    Use cases

    • Debugging an agent that consistently ignores specific instructions or behaves unpredictably.
    • Consolidating scattered rule files when migrating from .cursorrules to the newer .mdc format.
    • Cleaning up large, legacy instruction sets that have grown over 200 lines.
    • Setting up a shared, conflict-free rule standard for a multi-developer team.

    Known limitations

    Does not author brand-new rules or judge engineering quality. Limited to auditing persistent instruction files; does not audit transient chat history or one-off prompts.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.

    Only audits persistent instruction files. Cannot judge if a rule is good engineering practice, only if it conflicts or overlaps. Precedence logic varies by tool version.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 10 days ago

    Software
    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed10 days ago

    What's inside

    Frequently Asked Questions

    The auditor supports any AI tool that utilizes local instruction files, including Cursor (.cursorrules, .mdc), Claude Code (CLAUDE.md), Windsurf, and GitHub Copilot, as well as custom AGENTS.md files.

    The skill identifies direct contradictions where two rules command opposite behaviors, as well as 'scope-shadowing' where a global rule accidentally overrides a specific project-level instruction.

    By identifying and removing redundant or near-duplicate instructions, the auditor shrinks your rule files to their most potent form, saving valuable context window tokens and reducing 'instruction fatigue' for the AI.

    No, this auditor is strictly a diagnostic and optimization tool; it provides precise line numbers and ready-to-paste fixes so you remain in full control of your configuration.

    This is a one-time purchase that includes the current version of the auditor; updates are typically released when major AI agents introduce new rule formats or architectural changes.

    While it can handle any file size, this skill is most effective for teams with legacy rule sets or complex configurations exceeding 200 lines where manual auditing becomes error-prone.

    New in Code Quality & Review