Skip to content
| Marketplace
Sign in
Visual Studio Code>Programming Languages>PixelPilot Copilot: FiveM AI Agent, Copilot & AutocompleteNew to Visual Studio Code? Get it now.
PixelPilot Copilot: FiveM AI Agent, Copilot & Autocomplete

PixelPilot Copilot: FiveM AI Agent, Copilot & Autocomplete

PixelPilot AI Labs

|
305 installs
| (1) | Free
The AI agent for FiveM/GTA5 development: writes production-ready Lua (ESX, QBCore, ox_lib), runs your FXServer with one click and auto-fixes console errors until everything runs green. Inline autocomplete, stub protection, NUI/HTML generation, and 500+ AI models (Claude, GPT, Gemini, DeepSeek, Grok)
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

PixelPilot Copilot Discord Support Pricing

🚀 PixelPilot Copilot

Builds, runs and fixes your FiveM server. One click.

The AI agent built specifically for FiveM / GTA5 server development. Describe what you need — PixelPilot writes production-ready Lua, boots your FXServer and auto-fixes console errors until everything runs green.

No stubs. No placeholders. No -- TODO. And now: no alt-tabbing to a console window ever again.


▶️ Press play. It runs — and fixes itself. NEW

The headline feature of the latest release: a Run Server button right in the chat.

$ pixelpilot run-server
✓ FXServer found · server.cfg parsed — 42 resources
✓ ensure ox_lib … started
✓ ensure es_extended … started
✖ [script:esx_banking] SCRIPT ERROR: missing dependency 'oxmysql'
⚙ agent · reading stack trace…
⚙ agent · server.cfg fixed: ensure oxmysql before esx_banking
⚙ agent · restarting resource…
✓ ensure esx_banking … started
● 42/42 resources running · 0 errors
  • One-click run. The agent finds your FXServer and server.cfg anywhere in the project — even buried inside a start .bat — resolves the real launch command and boots it. The server keeps running in the background while you work.
  • Live console. Server output streams back to the agent, so it sees every error the moment it happens.
  • Auto-fix loop. Resource failed to start? The agent reads the stack trace, patches the broken code or server.cfg, restarts the resource and checks again — looping until the console is green. Hands-free.

💡 Open VS Code on your server's main folder (the one with the artifacts, server.cfg and the start .bat) so the agent can find everything.


✨ Why FiveM devs pick PixelPilot

  • 🎮 FiveM-native by default. Detects your stack (ESX, QBCore, qbox, ox_lib, oxmysql) on the first request and writes resources that match your conventions — fxmanifest.lua, client/server/NUI separation, real native calls that actually exist.
  • ⌨️ Inline autocomplete. Ghost-text completions tuned for Lua and FiveM patterns appear as you type — accept with Tab. Included on every paid plan.
  • 🛡️ Stub protection. Every file the agent writes is inspected before it touches disk. Near-empty client.lua? Blocked. Output with -- Existing code... or // TODO? Blocked and re-prompted until it's genuinely complete.
  • 🧠 Loop guards. Doom-loop detection, planning-loop interception, and auto-continue that resumes truncated writes from the exact next character.
  • 🔌 500+ AI models, one key. Sign in to the PixelPilot Gateway and route Claude, GPT, Gemini, DeepSeek, Grok and 500+ more through a single pp_ token — with live plan, quota and usage right in the editor. Or bring your own provider (OpenRouter, Anthropic, OpenAI, Google, Copilot, any OpenAI-compatible endpoint).
  • 📊 Usage in the status bar. Your 5-hour window, weekly window and prompt quota live in the status bar. Hitting a limit? Renew everything without leaving VS Code.
  • 🆓 Try before you buy. Three free prompts on a fresh install, no signup required.

🏎️ Quick start

  1. Install the extension from the VS Code Marketplace.
  2. Open your server's main folder — PixelPilot detects fxmanifest.lua and adapts to your stack.
  3. Sign in for full features: Profile icon in the sidebar → Sign in → your browser opens pixelpilotai.com/vscode-token → click 🚀 Send token to VS Code automatically.
  4. Type a prompt — "police job with duty system and MDT for ESX" — review the diffs, Accept.
  5. Hit ▶ Run Server and watch the agent boot FXServer and fix whatever breaks.

Don't want to sign in yet? You get 3 free prompts to try the FiveM Coder.


🤖 Built-in agents

Switch between specialized FiveM agents in the picker on the chat input.

Agent What it does
🟡 FiveM Coder (default) Ships complete resources end-to-end: fxmanifest.lua, client/server, NUI assets, locales. Enforces local scoping, CreateThread/Wait, correct source handling, ox_lib callbacks when present, and real natives — never placeholders.
🟣 FiveM Architect Plan first, code later. Resource structures, dependency graphs, database schemas, network sync patterns — outputs blueprints + a todo list for the Coder.
🔵 FiveM Debugger Hunts FiveM-specific failure modes: SQL injection, tight Wait(0) loops, unvalidated client events handing out money, NUI callbacks that never return, state replication bugs.

🛡️ Stub protection in detail

The single most-requested feature: an AI agent that doesn't ship skeleton code.

Every <write> is inspected before touching disk. Files below per-category minimum sizes, or containing any of the classic AI cop-outs (-- Existing code..., // rest of code, -- TODO implement later, // implement your logic here), are rejected — and the agent gets a focused re-prompt quoting the offending content back with a per-file checklist. Up to three attempts per file.

The result: when you accept a diff, the file is genuinely complete.


🔌 Providers & plans

  • PixelPilot Gateway (recommended) — one pp_ token, 500+ models, smart routing, server-side context compression, and your plan/quota/credits live in the Profile view. See plans →
  • Bring your own key — OpenRouter (with cross-model failover that rotates up to seven backups mid-stream), Anthropic, OpenAI, Google Gemini, Vercel AI Gateway, GitHub Copilot, or any OpenAI-compatible custom provider.

⚙️ Configuration

All settings live under pixelpilot-copilot.new.*:

Setting Purpose
apiToken Your pp_xxxxx token from pixelpilotai.com/vscode-token.
apiBaseUrl PixelPilot backend URL. Default https://pixelpilotai.com.
openRouterApiKey OpenRouter key for direct routing without the gateway.
autocomplete.enableAutoTrigger Inline ghost-text completions as you type.
claudeCodeCompat Load CLAUDE.md instructions + skills from your Claude Code config.
browserAutomation.enabled Playwright-powered browser automation tool.

Project-level config lives in .pixelpilot/ at the repo root: custom agents in .pixelpilot/agent/, commands in .pixelpilot/command/, MCP servers in .pixelpilot/pixelpilot.jsonc.


💬 Community & support

  • Discord — discord.gg/jG9b9MRaxM — the fastest way to get help, share scripts, and report bugs.
  • Support — pixelpilotai.com/support
  • Site, docs & pricing — pixelpilotai.com

📜 License

PixelPilot Copilot is proprietary software. © PixelPilot AI Labs. All rights reserved.

See LICENSE for the full text.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft