Code, preview, and ship
in one desktop IDE
Run Claude Code, Codex, or Grok Build with multi-model consults built in — then live-preview, publish, and buy a custom domain without leaving the app.
Feelin' Porki
Put your AI subscriptions on call inside Claude Code, Codex, or Grok Build.
The bundled MCP server exposes consult_claude_code, consult_codex, and consult_grok_build. Any of those three CLIs can now act as the master and ask either of the other two for a focused second opinion without spending PorkiCoder API credits or opening another window.
Bounded by design. Every call needs your explicit approval, receives only the question and context you choose, and runs read-only in a private temporary workspace. The worker cannot inspect your repository, edit files, run shell commands, browse, call MCP tools, or silently fall back to a paid API.
- Uses your existing Claude Code, ChatGPT/Codex, or Grok Build subscription login
- Consumes subscription quota, not PorkiCoder API credits
- Returns advisory markdown from only the bounded context you provide
- The
porkicoder-consultMCP server auto-registers with all three masters — no manual MCP configuration
Claude Sonnet 5 is the new Agent default.
New Agent sessions start on Claude Sonnet 5 at Max effort. Anthropic choices run through the Claude Agent SDK, while Gemini, Grok, and GPT choices automatically use the OpenCode harness with the selected PorkiCoder model carried into every turn.
The old pseudo-model named “OpenCode” is gone from the picker: choose the model you want and PorkiCoder chooses the harness. Existing Kimi, GLM, and legacy OpenCode preferences migrate safely to the new default; Kimi and GLM remain available to independent fan-out and HogCode workflows.
- Claude Sonnet 5 and Opus 4.8 support Low through Max effort
- Gemini 3.5 Flash, Gemini 3.1 Pro, Grok 4.3, and GPT-5.5 use OpenCode in Agent Mode
- Managed OpenCode defaults to Gemini and carries the chosen model explicitly
- Cmd+K inline edit now defaults to Gemini Flash, with Sonnet 5 optional
- Claude CLI discovery works from GUI launches and common version-manager installs
Built Different. Built Better.
Claude Sonnet 5 agents, working Gemini-named terminal tabs, local subscription consults, and a zero-config path from idea to live URL — in one IDE built from scratch.
Subscription Second Opinions, Inside Claude Code, Codex, and Grok Build
Run Claude Code, Codex, or Grok Build in PorkiCoder's terminal and it can ask either of the other two local subscriptions for a focused second opinion mid-task. The MCP server auto-registers, each worker stays isolated and read-only, and every call requires your approval — no repository access and no paid-API fallback.
Live Preview Server, Zero Config
Hit preview and PorkiCoder spins up a local server with live reload — auto-detecting Vite, your npm run dev script, or falling back to a static server. Every save in the IDE refreshes the page in milliseconds.
One-Click Publish + Buy a Domain
Ship to a free your-name.porkicoder.com subdomain in one click. Want your own domain? Search, buy, and connect a custom domain — DNS, SSL, and CDN provisioned automatically — without leaving the IDE.
Built From Scratch, Not a Fork
Not a VSCode fork. PorkiCoder is built from scratch on Electron + Monaco — every keystroke, every panel, every IPC channel ours. Bring your own API key for Claude / GPT / Gemini / Grok, pay providers directly, zero markup. A flat $20/month for the IDE — and use your existing local CLI subscriptions for second opinions without spending PorkiCoder API credits.
Four Core Model Families
Claude, GPT, Gemini, and Grok share one model picker. Switch models per message and let Agent Mode choose the matching Claude SDK or OpenCode harness. Kimi and GLM remain available in independent fan-out and HogCode workflows.
Agent Mode, Matched to Each Model
Agent Mode is the default: Claude models run through the Claude Agent SDK, while Gemini, Grok, and GPT run through OpenCode with the chosen model fixed explicitly. Drop into lightweight Chat Mode for a quick question.
Multi-Tab Terminal
A real node-pty terminal with multiple tabs, detach to a separate window, reattach later, and @terminal mentions to feed live shell output straight into chat as context.
Complete Conversation Control
Edit conversations mid-flow. Clear context instantly. Delete unwanted responses. @-mention any file from your project. Shape the AI's understanding in real time.
Privacy First
API keys stored encrypted locally via Electron safeStorage. Chat history saved only for your convenience. We never train any models on your code. Your IP stays yours.
From the Blog
Insights and analysis on the future of AI-powered development.
Mid-2026 Coding Tool Review: Agentic Desktops and Windsurf Updates
The AI Editor Wars Shift to the Desktop As we navigate through the summer of 2026, the landscape of AI coding assistants is shifting aggressively. For the past two years, the standard approach was to bolt an AI chat window onto existing text...
Read the full article →