Fast Grants for Agent Tooling on TON — Winners

What a hot few weeks it has been 👨‍💻

We received 40 strong submissions, and the depth, creativity, and technical ambition across them were genuinely impressive.

This was intentionally a fast, lightweight round designed to bootstrap experimentation and identify the strongest agent tooling ahead of a much larger Agent Competition.

It was so difficult to choose that we decided to expand the prize pool from $10K to $14K to support more projects.

In this round, we prioritized projects that provide reusable tooling, infrastructure primitives, and A2A building blocks that can accelerate the entire TON agent ecosystem.

14 winners, each receiving a $1,000 mini-grant:

Agent SDKs & Tools
• Teleton Agent by Resistor — autonomous AI agent platform for Telegram with native TON Blockchain integration
• AI sdk ton-tools by Mahdi — TON Blockchain tools for the Vercel AI SDK
• TON OpenClaw Skill by Michele Inetovsky — Claude Code / OpenClaw skill that gives AI agents full access to the TON Blockchain
• TON API Skill by Viacheslav Eliseev — Claude Code skill for TON API

Infrastructure & Security
• Ton AI Audit by One Last Phoenix — LLM-powered smart contract auditing for TON
• AgentGuard by Farzam — on-chain wallet guardrails for agents (spend limits, sessions, whitelists)
• TonPay 402 by Amir M. Mahmoudi — x402 protocol for machine-to-machine payments on TON
• x402-payments by Pavel Mulin — automatic payments for premium content and API access.
• Telehost by Allo — agent hosting platform
• TON Agent Platform by spend $ — no-code platform for building TON agents

Communication & Discovery
• TAMP by pwillis7 — on-chain agent registry and A2A discovery
• Mesh Network by Harjaap Dhillon — agent coordination protocol with on-chain reputation

Gifts Ecosystem
• Gift Asset MCP by Gift Asset API — structured access to Telegram Gifts & TON market data for agents
• Giftstat AI Bots by Camarade Ashik De Kano — Telegram Gifts analytics

Special Mention
• Many experts pointed out ton-analyst: agent skill to analyze TON data by Dan Okhlopkov. We liked it too. A lot. However, since Dan is part of the TON Foundation team, we decided to focus this round on external developers only. Dan, please forgive us 😄 You are building strong stuff, as always.

P.S. Not being selected this time does not mean your project is not valuable. Many strong submissions leaned more toward A2C applications rather than A2A infrastructure or agent tooling primitives, which was the primary focus of this round.

Keep building.

This fast round was just the beginning.

We are now preparing a much larger AI Agent Competition focused on the next layer of the stack.

Stay tuned 🚀