Prompt Lab

Prompt optimization loops and regression test suites for Claude Code, with a companion web UI.

The agent runs inside your Claude Code session and owns all LLM work — scoring responses, proposing improved prompts, applying suggestions. The server holds workspace state and keeps the agent and thePrompt Lab UIin sync.

{ "mcpServers": { "prompt-lab": { "type": "http", "url": "https://prompt-lab-mcp.up.railway.app/mcp" } } }

# 1. Open a workspace — agent shares the UI URL start_web_app() → "Open https://prompt-lab-mcp.vercel.app?s=abc123 to follow along." # 2. Register an API key register_api_key(workspaceId, "sk-ant-...") # 3. Set a system prompt and a test case set_system_prompt(workspaceId, "You are a concise customer support agent...") add_test_cases(workspaceId, [{ query: "How do I reset my password?", targetAnswer: "Click 'Forgot password' on the login page and follow the email link." }]) # 4. Run the optimization loop loop_optimization(workspaceId, threshold=85) → Iteration 1 — score 58: response too long, no mention of email link → Iteration 2 — score 74: better, but missing the exact step → Iteration 3 — score 91: SUCCESS — prompt updated to require step-by-step answers

The UI shows each iteration's score, the agent's reasoning, and the revised system prompt in real time.

Prompt Lab UI (github.com/jurek-f/prompt-lab) ↕ HTTP Prompt Lab MCP Server (Railway) ↕ MCP Claude Code (your machine)

API keys are never stored in the MCP server config. Instead, pass them to Claude Code as environment variables — the agent reads them and registers them with the server at the start of each session usingregister_api_key.

Set the key(s) for the provider(s) you want to use. The agent auto-detects the provider from the key prefix when callingregister_api_key.

If they're already in your system environment, Claude Code inherits them automatically — nothing else to do. Otherwise add them to~/.claude/envor your shell profile:

ANTHROPIC_API_KEY=sk-ant-... GEMINI_API_KEY=AIza... OPENAI_API_KEY=sk-...

Templates are global and appear in the UI dropdowns as soon as they are pushed.

Requires a workspace with at least one test case.

Deploy to Railway and set these environment variables:

npm install npm run dev # starts on :3000

This is a web browser that enables your coding agent, such as Claude Code, to visit websites on your behalf and assist you in identifying bugs or creating UI test cases.

Bring agent evaluations, observability, and synthetic test set generation directly into your IDE for free with Galileo's new MCP server

An MCP server to help AI assistants to answer questions and generate AccelByte Extend SDK code more effectively .

Instead of direct calling MCP tools, mcpcode server transforms MCP tool calls into TypeScript programs, enabling smarter, lower-latency orchestration by LLMs.

Help agents automatically write and test stories for your UI components

Supercharge your Agent with Semantic Code Intelligence and save 💰 in the process!

AI-to-AI code review platform — Claude, Codex, and Gemini cross-check each other via MCP, REST API, and CLI for consensus-based results.

A code sandbox for AI assistants to safely execute arbitrary code. Requires a 302AI API key for authentication.

Agent-native developer Q&A API with MCP + A2A endpoints for citations, job pickup, and answer submission.

MCP server to dynamically load Claude Code skills into AI agents

MCP bridge that lets Claude Code delegate heavy tasks to the Antigravity CLI (agy) — purpose-built tools, model routing with fallback, session continuity, and output truncation to save Claude's context and tokens.