Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    phouppo

    1

    Professional read-only audit for AI-generated apps covering security, architecture, data integrity, and HCI laws.

    Free

    Security scanned
    phouppo

    phouppo

    Example session with this skill installed

    /phouppo --depth standard --focus security,data

    • Read your context and instructions
    • Compiled the phouppo

    I have completed a standard audit of the repository focusing on security and data engineering. The scan identified 3 High-severity issues related to missing authorization checks and unindexed queries.

    Audit Report: Security & Data

    [!WARNING]
    Readiness Verdict: Ready with Conditions
    The application has critical architectural gaps in data isolation and potential IDOR vulnerabilities. Remediation is required before handling production user data.

    Severity Distribution

    • 🔴 Critical: 0
    • 🟠 High: 3
    • 🟡 Medium: 2
    • 🔵 Info: 4

    Findings

    <details open>
    <summary>🟠 <b>[SEC-001] Missing Row Level Security (RLS)</b> &bull; <code>High</code> &bull; <code>Effort: S</code></summary>
    
    PropertyValue
    Areasecurity
    Locationsupabase/migrations/20240101_init.sql:42
    ConfidenceHigh
    VerifiedStatic

    [!WARNING]

    Impact: Users can potentially read or modify data belonging to other tenants by manipulating UUIDs in API requests.

    Evidence:

    CREATE TABLE public.user_profiles (
      id uuid REFERENCES auth.users NOT NULL PRIMARY KEY,
      bio text,
      email text
    );
    -- Missing: ALTER TABLE user_profiles ENABLE ROW LEVEL SECURITY;
    

    Recommendation:
    Enable RLS on all public schema tables and implement policies using auth.uid().

    </details>
    
    <details open>
    <summary>🟠 <b>[DAT-001] Lost Update Race Condition in Balance Update</b> &bull; <code>High</code> &bull; <code>Effort: M</code></summary>
    
    PropertyValue
    Areadata
    Locationsrc/app/api/transfer/route.ts:18
    ConfidenceMedium
    VerifiedStatic

    [!WARNING]
    Impact: Concurrent transfers could result in incorrect account balances (Double Spend).

    Evidence:

    const { data: account } = await supabase.from('accounts').select('balance').eq('id', id).single();
    const newBalance = account.balance - amount;
    await supabase.from('accounts').update({ balance: newBalance }).eq('id', id);
    

    Recommendation:
    Use a database function (RPC) or db.rpc() to perform an atomic increment/decrement at the SQL level.

    </details>
    

    Next steps

    1. Review the full audit-report.md for the remediation checklist.
    2. Address the RLS findings in your next migration.
    3. Replace application-side arithmetic with atomic SQL updates for financial fields.

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

    About this skill

    Stop relying on the "happy path" of AI-generated code. Most vibe-coded applications fail at authorization boundaries, concurrency, and data invariants because the models that built them never say "I'm not sure." This skill provides a systematic, evidence-based audit of your codebase to find the critical engineering gaps before they become production incidents.

    What it does

    • Security Triage identifies exposed secrets, IDOR vulnerabilities, and broken server-side authorization.
    • Data Engineering Review checks for missing ACID transactions, unindexed foreign keys, and unsafe mass assignments.
    • Architectural Analysis flags N+1 query loops, serial waterfalls, and race conditions that cause tail latency.
    • HCI & UI Audit evaluates your frontend against Fitts's, Hick's, and Miller's laws to reduce cognitive load.
    • Interactive Reporting generates a full visual artifact with severity ratings, evidence snippets, and remediation checklists.

    How it works

    1. Reconnaissance identifies the stack, auth mechanisms, and database layers in under 10 tool calls.
    2. Phase 1 Triage runs 10 specific checks against foundational computer science principles to find immediate leaks or flaws.
    3. Phase 2 Deep Dive explores flagged areas like security, data, or visual design using scientific reference files.
    4. Verification runs read-only scanners or dynamic probes if enabled to confirm static findings.

    Frameworks & tools

    This tool is framework-agnostic but includes specialized logic for Next.js, React, Node.js, Python, Supabase, and SQL/NoSQL databases. It utilizes ripgrep for targeted searches and can integrate with existing testing suites.

    Why this beats prompting it yourself

    Generic prompts give you generic best practices. This skill uses a structured severity rubric and requires a "path:line" evidence trail for every finding, preventing the AI from hallucinating vulnerabilities or ignoring subtle logic bugs.

    Use cases

    • Post-Prompting Audit to review code generated by Cursor or Claude before merging.
    • Security Pre-flight to catch exposed keys or broken RLS before a production launch.
    • Performance Profiling to find event-loop blocking calls and database bottlenecks.
    • Due Diligence for quickly assessing the technical debt and risk in a "vibe-coded" prototype.

    Known limitations

    Does not perform write operations or fix code automatically. Static analysis of visual design is limited to CSS/component inspection unless dynamic mode is authorized.

    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 by design. Dynamic testing requires local execution consent. Cannot verify production-only environment variables or network-level firewall rules.

    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

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 12 days ago

    • Free with an Agensi account

    Software
    Node.jsPythonGit

    Permissions

    Terminal / Shell
    Read Files
    Network Access
    Write Files
    Listed12 days ago

    What's inside

    Frequently Asked Questions

    phouppo is designed to find critical engineering gaps—like broken authorization, N+1 queries, and data race conditions—that occur when AI models 'vibe-code' without considering deep architectural integrity.

    The skill is framework-agnostic but features optimized logic for Next.js, React, Node.js, Python, Supabase, and most SQL/NoSQL databases.

    No, this is a read-only audit skill designed to ensure safety; it identifies vulnerabilities and provides remediation checklists without ever modifying your source code.

    You receive the full skill package including the multi-phase audit logic, specialized reference files for HCI and security laws, and the interactive reporting template.

    phouppo requires read access to your codebase and utilizes tools like ripgrep for deep-file inspection; setup involves simply granting the skill permission to scan your local or cloud directory.

    By requiring 'path:line' evidence for every finding, phouppo forces the AI to ground its audit in your actual code, preventing the generic hallucinations common in standard prompting.

    New in Code Quality & Review

    Free