1. Applicant & Project Overview
Project Name: AgentID: The Decentralized Trust Registry for Autonomous Systems
Team / Individual Name(s): Gmin2 & Team
Links:@gmintu (in telegram)
\Summary:
AgentID is a decentralized “Yellow Pages” and reputation infrastructure for the Agentic Economy, built natively on the Intuition Protocol. It addresses the “trust gap” in multi-agent systems by allowing AI agents to register their identity (as Atoms), declare capabilities (as Triples), and accrue verifiable reputation (via Signal). Uniquely, AgentID implements “Bonded Identity,” requiring agent operators to stake $TRUST on their agents to deter malicious behavior. The project features a Model Context Protocol (MCP) server, enabling LLMs (like Claude or OpenAI) to query the Intuition Knowledge Graph in real-time to verify a counterparty agent’s trust score before transacting.
Project Category:
- Agent / Registry
- Identity / DID / Reputation
- AI Context / MCP
Elevator Pitch:
A reputation protocol where AI agents must “put their money where their code is”—using Intuition’s staking primitives to create a verifiable, Sybil-resistant credit score for the autonomous economy.
2. What You’re Building
Problem Statement:
We are entering the era of “Agentic Swarms,” where AI agents hire other agents to perform tasks (e.g., booking travel, executing DeFi swaps). However, currently, an agent has no way to verify if a counterparty agent is reliable, malicious, or prone to hallucinations. Centralized Web2 registries are siloed and easily manipulated, while current Web3 identity standards lack an economic cost for bad behavior.
Proposed Solution:
AgentID is a dApp and API that standardizes AI identity using Intuition primitives and aligns with the ERC-8004 (Trustless Agents) standard.
- Registry dApp: A UI where developers mint an Atom for their Agent and attach Triples defining its capabilities (e.g.,
[AgentX] --[canExecute]-->). - Bonded Reputation: A smart contract requirement that forces Agent developers to stake Signal ($TRUST) on their Agent’s Atom. If the Agent performs poorly, the community counter-signals (slashes) this stake.
- The “TrustCheck” MCP Server: An MCP integration that allows major AI models to query the Intuition graph (e.g., “Is Agent X reliable?”) and receive a cryptoeconomically weighted answer.
Stage of Development: Idea / Prototype (Architecture defined, schema ready for Testnet).
Technical Architecture Overview:
- Frontend: Next.js +
@0xintuition/1uifor the registry interface. - Backend: A custom TypeScript MCP Server that queries the Intuition Rust Subnet via GraphQL.
- Protocol: Direct integration with
EthMultiVaultcontracts on the Base L3. - Standards: Aligns with ERC-8004 for agent identity and x402 for future agent-to-agent payments.
3. Team & Execution Ability
Team Members:
- Gmin2 - Lead Developer & Founder
- ItsMoh - Blockchain Wizard & Co-Founder
Relevant Expertise:
- Gmin2 have experience with FFI and have a rust crate that deal with all the FFI bindings of Cups-rs (linux official printing system) and the rust crates that deal with the printing system.
- We have previously completed the InkBench project which was funded by the polkadot ecosystem which was purely written in rust which is a in built benchmarking tool for the ink! smart contract language.
- We have also worked on developing an interactive ctf game where we teach common smart contract attacks and exploits
- I have the privileged of working with the official cargo team in implementing the autocompletion of the cargo command line tool
See all my works in (Gmin2) (github) (PS: I was restricted to posting only 2 links)
Execution Proof:
InkCTF! · GitHub
GitHub - ItshMoh/inkBench
Commitment Level: Full-time for the duration of the grant (3 months).
4. Grant Request & Milestones
Amount Requested: $28,000 (Payable in USDC)
Budget Breakdown & Milestones:
| Milestone | Timeline | Deliverables & Success Criteria | Budget Allocation |
|---|---|---|---|
| M1: Schema & Core Registry | Weeks 1-4 | Deliverables: • JSON-LD Schema for Agents (aligned w/ ERC-8004). • CLI tool to mint Agent Atoms & Capability Triples. Success Criteria: • Successful minting of 10+ “Agent Atoms” on Intuition Testnet. • Verified query of capabilities via GraphQL. |
$8,000 |
| M2: Bonding UI & Dashboard | Weeks 5-8 | Deliverables: • Web Dashboard (Next.js) for searching agents. • Staking/Unstaking UI using @0xintuition/1ui.Success Criteria: • Users can connect wallet, search an Agent Atom, and deposit $TRUST to increase Signal. • “Trust Score” visualizer is live. |
$10,000 |
| M3: MCP Server & SDK | Weeks 9-12 | Deliverables: • Open-source MCP Server (TypeScript). • get_agent_reputation tool exposed to LLMs.Success Criteria: • Demo video of Claude/ChatGPT deciding to trust/distrust an agent based on Intuition data. • 5+ External Agents registered. |
$10,000 |
How this compounds value:
This project creates the fundamental “phonebook” for the AI ecosystem on Intuition. By enabling agents to verify each other, we unlock high-frequency automated transactions, significantly increasing network activity and knowledge density.
5. Intuition Ecosystem Alignment
Why Intuition:
Intuition is the only protocol that combines Identity (Atoms) with Economic Weight (Signal). A simple registry (like ENS) is insufficient for AI agents because it lacks a mechanism to punish bad actors. Intuition’s bonding curves allow us to create a dynamic “Credit Score” that fluctuates based on real-time performance and community consensus.
Intuition Primitives Used:
- Atoms: To represent the AI Agents (DIDs) and Model types.
- Triples: To map capabilities (
[Agent] --[supportsModel]--> [Llama3]) and endorsements ([User] --[trusts]--> [Agent]). - Signal: The core metric for the “Trust Score.” High signal = High collateral = Trusted Agent.
- MCP: Implementing the Model Context Protocol to make this data natively consumable by LLMs.
Increasing Knowledge Density:
By forcing agents to bond capital to their identity, we filter out “spam” agents. Only agents with high conviction (Signal) rise to the top of the registry, ensuring the graph remains high-signal and distinct from low-cost Web2 databases.
6. Sustainability & Long-Term Vision
Long-Term Vision (6-24 Months):
To become the “SSL Certificate” of the AI age. Just as you wouldn’t enter a credit card on a website without HTTPS, no AI agent should transact with another agent without checking its AgentID verification first. We aim to integrate x402 payments so agents can pay for this verification autonomously.
Post-Grant Sustainability:
- Verification Fees: We will charge a small fee (in $TRUST) for “Verified Badge” issuance (e.g., verifying codebase matches on-chain hash).
- Curation Rewards: As the creators of the “Agent Registry” Triples, we will earn a percentage of the curation fees as the ecosystem grows and more users signal on our curated lists.
- B2B API: Premium API access for high-frequency trading bots to query reputation scores with lower latency.
7. Additional Materials
System Architecture & User Flow:
Below is the technical flow for how an Agent registers and is verified by a client.
User Flow Description:
- Registration: Developer mints an Agent Atom on the Intuition L3 via our UI.
- Bonding: Developer stakes $TRUST tokens to signal honest intent and bootstrap the agent’s reputation.
- Verification: A Client Agent (via our MCP Server) queries the Intuition Graph for Signal strength.
- Execution: If the Signal meets the threshold, the Client Agent proceeds with the task/payment.
8. Applicant Attestation
I confirm that all information submitted is accurate and that I intend to deliver the listed milestones.
Name: Gmin2
Wallet Address: 0x7d26625133a2964d133dcdf228da319877557ec1
Date: 5/12/2025