Build with AI

AgentTransfer gives your AI agents their own identity — a name, file folder, inbox, and email address — with one API call. Use it to transfer files up to 5 GB between agents over expiring, sha256-verified links, with every action logged in a signed, tamper-proof receipt. A built-in MCP server makes it easy to plug into any agent framework.

Scribe is a command-line tool that reads your git history and AI coding sessions (Claude Code, Codex), then automatically writes a structured wiki you can use as memory for your AI agents. Instead of spending time maintaining notes, you install one binary, let it run on a schedule, and your agents pick up your decisions and context before every session — all stored as plain markdown in a git repo you own.

Open Vibe turns your AI coding agent — like Claude Code — into a step-by-step SaaS-building tutor. Instead of prompting blindly and burning credits on broken loops, you get guided lessons with live diagrams that show what actually happens inside your running app. You ship a working SaaS and understand every part of it.

PrePrompt is an open-source tool that sits between you and your LLM inside Claude Code and Cursor. It intercepts every prompt you type, scores its clarity in under 1ms using local Python heuristics, and rewrites vague requests into detailed technical specs before they reach the model. Over time it learns your tech stack — framework, language, database, style — and injects that context automatically into every optimization without you having to repeat yourself.

PollyReach is an MCP skill that gives any AI agent — Claude, ChatGPT, or any MCP-compatible assistant — a dedicated real phone number and the ability to make live phone calls in 50+ countries. Your agent books restaurant reservations, schedules appointments, handles customer support calls, or qualifies leads by actually dialing the number. You get a monthly free call quota; usage beyond that is pay-as-you-go with no upfront payment.

Buda lets you run a full team of AI agents that handle sales, marketing, operations, and coding — all from one dashboard. Your agents get persistent memory through Buda Drive, so they remember past tasks and keep working without starting over. You can deploy them to Slack, Discord, Telegram, and more, so your team gets help wherever they already work.

Coasty gives you a REST API that runs autonomous tasks on managed Linux and Windows machines, so your app can delegate browser work to a computer-use agent without owning the infrastructure. You describe a goal, Coasty drives the machine to completion, and you watch progress through live SSE events or signed webhooks. It's a YC S26 company and bills $0.05 per completed step.

OpenClaw Launch gives you a personal AI agent running on a dedicated cloud machine—no Mac Mini to buy, no Docker to configure, no command line to touch. You click Deploy, and in 30 seconds your agent is live with a full browser, terminal, and file manager. It connects to Gmail, Slack, GitHub, Google Drive, Google Calendar, and 1,000+ more apps without you entering a single API key.

ask-a-human is a free, open-source MCP server that gives your AI agents a way to ping your phone when they hit a decision they shouldn't make alone. You paste one config snippet into your agent's mcp.json — no account, no API key required. When an agent gets stuck, it sends an end-to-end encrypted message to your phone and waits for you to approve, decline, or type a quick reply before continuing.

Vibedock sits in your macOS menu bar and lets you toggle Claude Code MCP (Model Context Protocol) servers on and off with a single click — no JSON config editing needed. It automatically kills and relaunches your Claude sessions after each toggle, tracks how many tokens you save by disabling idle servers, and stays organized by project so you always know which tools are active.