home / skills / yuniorglez / gemini-elite-core / ai-pro

ai-pro skill

/skills/ai-pro

This skill orchestrates autonomous agents and enterprise-grade tooling for secure, high-performance AI systems using GPT-5, Auth.js v5, and Stripe v13.

npx playbooks add skill yuniorglez/gemini-elite-core --skill ai-pro

Review the files below or copy the command above to add this skill to your agents.

Files (13)
SKILL.md
6.6 KB
---
name: ai-pro
description: Professional AI and System Orchestrator. Specialized in GPT-5, autonomous agents, Auth.js v5, Stripe v13, Monorepos (Bun/pnpm), and large codebase intelligence. Optimized for January 2026 standards.
---

# AI-PRO: Professional AI & System Orchestrator

## Overview
**AI-PRO** is the definitive skill for managing modern, high-performance software ecosystems in 2026. It bridges the gap between raw AI capabilities (GPT-5, o3-deep-research) and enterprise-grade infrastructure (Auth.js v5, Stripe v13, Bun/pnpm Monorepos).

This skill is designed for the **Senior AI Engineer** who needs to orchestrate autonomous agents while maintaining strict security, performance, and architectural integrity.

## Core Pillars of the 2026 Stack

### 1. The AI Frontier (GPT-5 & Autonomous Agents)
The paradigm has shifted from "Chat" to "Agents". We utilize the GPT-5 family for complex tool-use and `o3-deep-research` for iterative discovery missions.
- **Reference**: [`references/ai-agents.md`](./references/ai-agents.md)

### 2. Identity & Security (Auth.js v5)
Security in 2026 is "Edge-First". Auth.js v5 provides 25% faster validation and native Edge support using the standardized `AUTH_` prefix.
- **Reference**: [`references/auth-v5.md`](./references/auth-v5.md)

### 3. Financial Infrastructure (Stripe v13+)
Modern API consumption leverages auto-paginating SDKs and deeply expanded objects to reduce network overhead and boilerplate.
- **Reference**: [`references/api-v13.md`](./references/api-v13.md)

### 4. Codebase Intelligence (Architect & Archive)
Managing massive codebases requires "Context Packing" with Repomix and high-performance search via `git grep` and indexed symbol maps.
- **Reference**: [`references/architect-archive.md`](./references/architect-archive.md)

---

## Strategic Implementation Guide

### The "Autonomous Orchestrator" Pattern
To build a system that can self-heal and evolve, follow the **O.R.C.A.** (Observe, Reason, Check, Act) pattern.

#### Example: Self-Healing API Integration
```typescript
import { gpt5 } from '@openai/provider';
import { generateText } from 'ai-sdk';
import { auth } from '@/auth';
import Stripe from 'stripe';

/**
 * AI-PRO Orchestrator: Automatically fixes API version mismatches
 */
async function selfHealingPayment(customerId: string) {
  const session = await auth(); // Auth.js v5 Edge-First
  const stripe = new Stripe(process.env.AUTH_STRIPE_SECRET!, {
    apiVersion: '2025-10-01',
  });

  try {
    const customer = await stripe.customers.retrieve(customerId, {
      expand: ['subscriptions.data.default_payment_method'],
    });
    // Business logic...
  } catch (error) {
    // If error is related to API versioning or schema change
    const diagnostic = await generateText({
      model: gpt5('gpt-5-pro'),
      prompt: `Analyze this Stripe error: ${error.message}. 
               Context: We are using SDK v13. 
               Suggest a fix or retry strategy.`,
    });
    
    console.log(`AI-PRO Diagnostic: ${diagnostic.text}`);
    // Implement auto-retry or fallback logic here
  }
}
```

---

## Advanced Search & Context Management

### Ripgrep Masterclass
When searching through million-line archives, speed is safety.

| Command | Use Case |
| :--- | :--- |
| `rg -l "pattern"` | List ONLY filenames containing the pattern. |
| `rg -C 5 "pattern"` | Show 5 lines of context around the match. |
| `rg -t typescript "pattern"` | Search ONLY in TypeScript files. |
| `rg --stats "pattern"` | Show performance stats (count, time). |

### Repomix Strategy
Before starting a complex refactor, generate a "Codebase Map":
```bash
bun x repomix --include "src/core/**,src/types/**" --output .ai-context/core-map.xml
```

---

## The "Do Not" List (Anti-Patterns to Avoid)

### ❌ AI & Agents
- **Do NOT** allow an agent to commit to `main` without a `strict-auditor` check.
- **Do NOT** hardcode model names; use a registry or environment variables to allow for easy upgrades (e.g., GPT-5 -> GPT-6).
- **Do NOT** use agents for deterministic tasks (like regex) where a simple script suffices.

### ❌ Authentication
- **Do NOT** use `process.env.NEXTAUTH_SECRET`; v5 expects `AUTH_SECRET`.
- **Do NOT** store PII (Personally Identifiable Information) in JWTs; keep tokens lean for Edge performance.
- **Do NOT** ignore the `AUTH_TRUST_HOST` requirement in non-Vercel environments.

### ❌ API & SDKs
- **Do NOT** manually handle Stripe pagination using `starting_after`; it is error-prone. Use `for await...of`.
- **Do NOT** neglect webhook signature verification; it is the #1 vector for payment fraud.

### ❌ Architecture
- **Do NOT** create "Monster Monorepos" without clear package boundaries. Use `pnpm-workspace.yaml`.
- **Do NOT** include `node_modules` in your `repomix` packs; it will blow your token budget.

---

## Evolution Roadmap (2026-2027)

### Q1 2026: The o3 Stabilization
- Deep integration of `o3-deep-research` into CI/CD pipelines for autonomous bug hunting.
- Standardization of the `AUTH_` prefix across all Squadds projects.

### Q2 2026: GPT-5 Multi-Modal Agents
- Native visual auditing: Agents that "see" the UI to find CSS regressions.
- Real-time audio debugging: Voice-driven system diagnostics.

### H2 2026: The GPT-6 Horizon
- Early testing of "Infinite Context" models.
- Transition from Monorepos to "Micro-Repos" managed by a Centralized AI Architect.

---

## Quick Reference Commands

### Identity Setup (Auth.js v5)
```bash
# Generate AUTH_SECRET
openssl rand -base64 32
```

### Dependency Sync (Bun + pnpm)
```bash
# Reinstall all workspace deps with Bun's speed
bun install --frozen-lockfile
```

### Context Packing
```bash
# Pack core logic for AI review
bun x repomix --style markdown --output review.md
```

### Large Scale Search
```bash
# Find all sensitive keys across the archive
git grep -E "(SECRET|KEY|PASSWORD)" -- ':!*.lockb'
```

---

## Troubleshooting

### Issue: "Missing AUTH_SECRET"
- **Cause**: Auth.js v5 requires this for encryption in production.
- **Fix**: Add `AUTH_SECRET` to your `.env` or CI secrets.

### Issue: "Stripe SDK Pagination Error"
- **Cause**: Trying to access `.data` on an async iterator.
- **Fix**: Use `for await (const item of stripe.customers.list())` directly.

### Issue: "Repomix Token Overflow"
- **Cause**: Packing too many files or large assets.
- **Fix**: Use `--ignore` or create a specific `repomix.config.json` for the module.

---

## Metadata
- **Version**: 2.0.0 (January 2026 Refactor)
- **Primary Agent**: AI-PRO
- **Sub-Agents**: `auth-expert`, `api-pro`, `architect-pro`, `archive-searcher`
- **Compatibility**: Next.js 16.1+, React 19.2+, Stripe 13.5+, Bun 1.2+

*Updated: January 22, 2026 - 15:20*

Overview

This skill is a professional AI and system orchestrator optimized for January 2026 stacks. It combines advanced agent orchestration (GPT-5 and autonomous agents), identity/security with Auth.js v5, payment integration with Stripe v13, and large-codebase intelligence for monorepos and archive search. It targets senior AI engineers who need reliable, secure, and high-performance automation across CI/CD and production systems.

How this skill works

The skill provides patterns, utilities, and protocols to observe system state, reason about errors, check constraints, and act (O.R.C.A.). It inspects API errors, authentication configuration, and codebase context to generate diagnostics, automated retries, or remediation actions. It also supplies high-speed search and context-packing tooling (ripgrep, repomix) to construct compact, relevant context for agent decision-making.

When to use it

  • Automating self-healing integrations with external APIs (Stripe, etc.).
  • Orchestrating autonomous agents that require strict security and audit controls.
  • Packing and supplying compact code context to agents for large monorepos.
  • Performing high-speed codebase search and symbol mapping across millions of lines.
  • Integrating Edge-first authentication flows using Auth.js v5 in modern deployments.

Best practices

  • Use environment-driven model registries rather than hardcoding model names to simplify upgrades.
  • Enforce strict-auditor checks before agents commit to main branches or deploy changes.
  • Prefer async iterators for SDK pagination (for await...of) instead of manual starting_after handling.
  • Keep JWTs lean; avoid storing PII and use AUTH_ prefixed secrets for v5 deployments.
  • Create targeted repomix packs with explicit include/ignore lists to limit token usage.

Example use cases

  • Auto-diagnose and remediate Stripe API version or schema errors with GPT-5 generated diagnostics and retry logic.
  • Edge-first Auth.js v5 setup that validates AUTH_SECRET and enforces AUTH_TRUST_HOST for non-hosted environments.
  • Generate a codebase map with repomix and ripgrep to produce a compact AI review bundle for a large refactor.
  • Run automated agent audits that prevent unsafe changes to critical packages in a pnpm/Bun monorepo.
  • Use high-performance search commands to find sensitive keys and enforce secret-remediation workflows before CI artifacts are published.

FAQ

What does O.R.C.A. stand for and why use it?

O.R.C.A. is Observe, Reason, Check, Act. It structures autonomous workflows so agents first gather state, then reason with models, validate decisions, and perform safe actions with audit trails.

How do I avoid repomix token overflows?

Limit included paths, use --ignore for large assets, and build dedicated repomix configs that exclude node_modules and binary blobs.

Which secret names are required for Auth.js v5?

Use AUTH_SECRET as the primary encryption secret and adopt the AUTH_ prefix convention for related environment keys to ensure Edge-first compatibility.