Yadkin LabsYadkin Labs

Open Source

Projects

Everything we build is open source. Real systems we use ourselves. MIT licensed. Fork it, deploy it, build on it.

🎭

yadkinlabs-meme-maker

Agent-native video meme API

An agent-native REST API that generates video memes end-to-end. Agents submit a prompt and optional base image, receive a price offer over x402, pay in USDC on Base, and get a rendered video back — no API key, no account, no UI. Deployed at meme-maker.yadkinlabs.ai.

  • x402 payment — agents pay per job in USDC, no API key needed
  • Claude for script, gpt-image-1 for footage, ElevenLabs for voice
  • Runway Gen-4.5 for motion, ffmpeg for final assembly
  • Trigger.dev durable tasks survive timeout and cold starts
  • Built on agentic-gateway + agentic-content

Video coming soon

agentic-gateway

Cloudflare Workers edge layer for agentic APIs

A drop-in Cloudflare Workers proxy that sits between AI agents and your backend. Handles x402 payment verification (USDC on Base), WAF (oversized payloads, prompt injection), and audit logging — so your backend stays clean and focused on its job.

  • x402 payment verification via x402.org facilitator
  • WAF middleware blocks oversized payloads and prompt injection
  • Audit log forwarded to any HTTP sink
  • Deploys to Cloudflare Workers — global edge, zero cold starts
  • Shared across all products; per-product config via env vars

Video coming soon

🧩

agentic-content

Reference implementation for AI content pipelines

A Next.js + Trigger.dev reference implementation for building agent-native content APIs. MCP server, durable pipeline tasks, Cloudflare R2 storage, and Prisma + Neon — the complete stack for building products where agents are first-class users.

  • Trigger.dev v3 durable tasks — survive cold starts, long timeouts
  • MCP server for agent tool access
  • Cloudflare R2 for artifact storage (audio, video, images)
  • Prisma + Neon PostgreSQL — pooled connections for serverless
  • Reference implementation — fork and adapt per product

Video coming soon

🎬

agentic-video

AI video production pipeline

A full video production system driven by AI agents — from topic research through scripting, voiceover, image generation, motion, assembly, and YouTube upload. Every step has a human review gate. Built to run on a Raspberry Pi or any Linux box.

  • End-to-end pipeline from topic to published video
  • Human review gates at every major step
  • Multi-channel support with per-channel config
  • Claude for research, scripting, and smart fix
  • ElevenLabs, Runway, and configurable image gen

Video coming soon

🤖

agentic-oss

AI triage for open source repositories

A GitHub Action that automatically triages issues and PRs using Claude. Questions get answered and closed. Duplicates get linked and closed. Spam gets closed silently. Bugs and feature requests get labeled and queued for you. One daily digest, not dozens of notifications.

  • Handles questions, duplicates, spam, and stale issues automatically
  • Daily stale issue management
  • Daily digest issue with AI summary
  • Designed to resist prompt injection from issue and PR content
  • Zero infrastructure — pure GitHub Actions

Video coming soon

More in development. Follow us on GitHub to stay updated.