MoltLaunch — Trust Infrastructure for AI Agents

Composable trust signals for AI agents on Solana.
We verify infrastructure. Protocols decide what "trusted" means.

---

The Problem

Creating 1,000 fake agent identities on Solana costs $0.

Protocols can't distinguish verified agents from spam. On-chain anchoring proves data existed — not that it's true. The gap is verification: who checked, how they checked, why you should believe them.

The Solution

Composable trust signals, not trust scores.

AgentIdentity PDA stores raw attributes (infra_type, economic_stake, hardware_binding). Protocols compose their own access policies from these signals.

We provide signals. Protocols decide.

---

Architecture (V3.1)

Four PDAs, twelve instructions. Everything derives from seeds:

PDA	Seeds	Purpose
ProtocolConfig	["moltlaunch"]	Singleton config, revocation nonce
Authority	["authority", pubkey]	Per authorized verifier
AgentIdentity	["agent", wallet]	The signal hub — raw trust attributes
Attestation	["attestation", wallet, authority]	Per verification receipt

12 instructions. Permissionless refresh. Stale-on-Read pattern — attestations carry timestamps; consumers decide freshness thresholds at read time, not write time.

---

Quick Start

Register and verify an agent via the self-verify API at proveyour.id:

# Register agent
curl -X POST https://proveyour.id/api/self-verify \
  -H "Content-Type: application/json" \
  -d '{"agentId":"my-agent","acceptTerms":true,"name":"My Agent"}'

# Check status
curl https://proveyour.id/api/self-verify/agent/my-agent

---

On-Chain Program

Program: 6AZSAhq4iJTwCfGEVssoa1p3GnBqGkbcQ1iDdP1U1pSb (devnet)
Tests:   43/43 passing

Source: github.com/tradingstarllc/moltlaunch

---

SDK

One package. @moltlaunch/sdk v3.0.0

npm install @moltlaunch/sdk

import { MoltLaunchSDK } from "@moltlaunch/sdk";

const sdk = new MoltLaunchSDK({
  endpoint: "https://proveyour.id",
});

// Register an agent identity
const identity = await sdk.registerAgent({
  agentId: "my-agent",
  name: "My Agent",
  acceptTerms: true,
});

// Fetch trust signals
const signals = await sdk.getAgentSignals("my-agent");
console.log(signals.infraType);       // "cloud"
console.log(signals.economicStake);   // 1.5 SOL
console.log(signals.hardwareBound);   // true

---

Composability

MoltLaunch is designed to compose with the emerging Solana agent ecosystem:

Protocol	Integration	Pattern
AAP (Agent Authority Protocol)	Delegation scopes	Two-CPI pattern — AAP checks delegation, MoltLaunch checks identity
SATI (Solana Agent Trust Interface)	Reputation feedback	SATI reads MoltLaunch attestations as reputation inputs
Solana ID	Human verification	Complementary — Solana ID verifies humans, MoltLaunch verifies agents
SAS (Solana Attestation Service)	Storage migration	Attestation data portable to SAS format for cross-protocol consumption

The two-CPI pattern with AAP: a protocol calls AAP to verify "is this agent authorized to act?", then calls MoltLaunch to verify "is this agent real?" Both must pass.

---

Ecosystem Position

Layer 1 in the 7-layer Solana Agent Stack — the identity and trust foundation that other layers build on.

See the full composability map: youragent.id/blog.html

---

Links

• 🌐 Website: youragent.id
• 🔐 Self-Verify: proveyour.id
• 🎮 Demo: youragent.id/demo.html
• 📝 Blog: youragent.id/blog.html
• 🚢 Shipped: youragent.id/shipped.html
• 📦 SDK: npm install @moltlaunch/sdk
• 📋 sRFC #9: github.com/solana-foundation/SRFCs/discussions/9
• ⚓ Anchor Program: github.com/tradingstarllc/moltlaunch

---

Hackathon Journey

Four architecture iterations: V1 (verification API) → V2 (STARK proofs) → V3 (composable signals) → V3.1 (Stale-on-Read + permissionless refresh).

Each pivot included a public self-critique. 90+ forum posts. 311 projects mapped across the hackathon ecosystem. Most detailed technical thread in hackathon history.

---

License

MIT © 2026 TradingStar LLC

---

Built for the Colosseum Agent Hackathon 2026