Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    codebase improvement advisor

    1

    Senior-level codebase auditor that generates evidence-backed improvement plans and prioritized technical roadmaps.

    $12

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    codebase improvement advisor

    codebase improvement advisor

    Example session with this skill installed

    Run a standard audit of the current repository. Focus on high-leverage tech debt and performance bottlenecks. Generate plans for the top three findings.

    • Read your context and instructions
    • Compiled the codebase improvement advisor

    I have completed a standard audit of the repository. I analyzed the directory structure, package manifests, and CI configurations to identify high-leverage improvements. I have vetted all findings against source evidence and prioritized them by impact and confidence.

    Codebase Audit Results

    IDTitleCategoryImpactEffortRiskConfidence
    PERF-01N+1 Query in Dashboard LoaderPerformanceHighMLowHigh
    DEBT-01Circular Dependency in Auth ModuleTech DebtMedSMedHigh
    TEST-02Missing Integration Tests for Payment FlowTest CoverageHighLLowHigh
    SEC-01Unvalidated Input in Search APISecurityCriticalSLowHigh

    [PERF-01] N+1 Query in Dashboard Loader

    • Evidence: src/api/dashboard.ts:42 - The getUserStats call is inside a map function over the workspace list.
    • Impact: Significant latency increase for users with more than 5 workspaces.
    • Effort: M.
    • Risk: LOW - standard refactor to join query.
    • Confidence: HIGH.

    Fix sketch: Refactor the data fetcher to use a single batch query for all workspace IDs instead of individual calls.

    [SEC-01] Unvalidated Input in Search API

    • Evidence: src/controllers/search.ts:18 - Query parameter q is passed directly to the database driver.
    • Impact: Potential SQL injection risk.
    • Effort: S.
    • Risk: LOW - fix involves adding a validation schema.
    • Confidence: HIGH.
    • Fix sketch: Implement Zod validation for all incoming search query parameters.

    [DEBT-01] Circular Dependency in Auth Module

    • Evidence: src/auth/service.ts imports from src/user/service.ts, which imports back from src/auth/utils.ts.
    • Impact: Increases build times and prevents clean modularization.
    • Effort: S.
    • Risk: MED - requires careful re-ordering of exports.
    • Confidence: HIGH.
    • Fix sketch: Move shared types to a new src/common/types/auth.ts file to break the loop.

    Next steps

    • I have created detailed execution plans for PERF-01, SEC-01, and DEBT-01 in the plans/ directory.
    • Review plans/README.md for the suggested execution order.
    • You can now hand these plans to an executor agent or a developer.

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

    About this skill

    The problem

    Large codebases accumulate tech debt and architectural drift that manual reviews often miss. Engineering teams struggle to turn vague audit findings into actionable, self-contained implementation plans that other developers or agents can actually execute.

    What it does

    • Performs multi-category audits covering security, performance, test coverage, and developer experience.
    • Vets findings against concrete file evidence to eliminate false positives and "hallucinated" bugs.
    • Prioritizes work by leverage, calculating impact-to-effort ratios for every identified issue.
    • Generates self-contained, machine-executable implementation plans with clear stop conditions and verification gates.
    • Reviews existing pull requests or branches to identify introduced risks before they merge.

    Frameworks & tools

    Works with any language or framework by analyzing root manifests, CI configs, and directory structures. Specifically leverages plans/ or advisor-plans/ directories for artifact storage.

    Why this beats prompting it yourself

    Most LLM audits provide "vibes" and generic advice. This skill enforces a rigorous evidence-backed protocol, requiring the agent to open and confirm every file and line number cited. It separates the advisor role from the executor role, ensuring plans are robust enough to stand alone without session context.

    Use cases

    • Conducting deep security and performance audits on legacy repositories.
    • Onboarding new agents or engineers by creating a verified roadmap of high-leverage fixes.
    • Performing pre-merge risk assessments on complex feature branches.
    • Standardizing tech debt documentation into executable handoff templates.

    Known limitations

    This skill is strictly read-only for source code. It will not implement fixes, commit changes, or run mutating commands in your working tree.

    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.

    Read-only for source code. Requires repository access. Cannot implement fixes or run mutating commands. Execution requires separate subagent/worktree support.

    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 2 months ago

    Software
    Git

    Permissions

    No special permissions declared or detected

    Listed2 months ago
    Updated1 month ago

    What's inside

    Frequently Asked Questions

    This skill acts as a senior-level auditor that scans your entire repository to identify architectural drift and technical debt, turning vague issues into structured, evidence-backed implementation plans for your team or other AI agents to execute.

    The advisor is language-agnostic; it identifies frameworks and patterns by analyzing root manifests and directory structures, making it compatible with any codebase regardless of the primary programming language.

    No, this skill is strictly read-only to ensure safety; it generates the roadmap and verification gates, but it will never commit code or mutate your source files directly.

    The advisor is designed for use with advanced LLM-based agents that have file-system access, and it outputs its findings into standardized `plans/` or `advisor-plans/` directories for easy integration into your workflow.

    Unlike generic tools, this skill requires the agent to verify every finding against specific file evidence and line numbers, eliminating "hallucinations" and ensuring every recommendation is grounded in your actual code.

    You receive the full skill configuration, including the proprietary evidence-backed auditing protocols and structured output templates for generating machine-executable implementation plans.

    New in Code Quality & Review