Oracle3
Open-source trading engine and MCP server for prediction markets. Fee-aware no-arbitrage trading across related contracts, live on Kalshi, Polymarket and Solana or on paper.
An autonomous risk desk for on-chain and tokenized-asset portfolios. It grades your risk, hires specialist agents to fix it, pays them on-chain, and settles on X Layer. You hold the kill switch.
FinanceTypeScript · JavaScriptGitHub @Excellency001-boopOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
No updates detected yet. GitHub releases will appear here.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
Second Opinion Pro — hosted, always-on autonomous risk desk for OKX Dev Day 2026. Grades your portfolio, hires agents to fix it, settles on X Layer. You hold the kill switch.
Taken from the repository README
Loading the full README. The saved excerpt is shown below.
The risk desk that acts. An autonomous risk desk for on-chain and tokenized-asset
portfolios, built for OKX Dev Day 2026. It grades your portfolio across seven
vitals (A+ to D), then a coordinator agent hires specialist agents (A2A), pays them
per task (x402), and has them execute rebalances on X Layer. You hold the
guardrails and the kill switch. Every action is logged with on-chain references.
Spans both tracks: OKX AI (a real multi-agent economy with discovery,
coordination and monetisation) and X Layer (tokenized stocks / RWA put to work
in an automated strategy).Read the full README ↗Most tools tell you your crypto portfolio is risky and stop there. Second Opinion Pro grades the risk, then hires and pays small AI specialist agents to actually fix it on X Layer, while you hold the limits and the kill switch the whole time.
▶ Live demo: https://second-opinion-pro.vercel.app Built for: OKX Dev Day 2026 · Tracks: OKX AI (primary) + X Layer / RWA
Try it in 20 seconds: open the link, keep "Demo · at-risk book", press Get a second opinion, then Approve plan, and watch the grade climb from D to A− while you hold a red kill switch.
flowchart LR
P[Your portfolio] --> G{Grade A+ to D}
G -->|risky| PLAN[Draft a fix plan]
PLAN --> A[You approve: caps and kill switch]
A --> H[Coordinator hires agents via A2A]
H --> PAY[Pays them per task via x402]
PAY --> X[Execute on X Layer]
X --> R[Re-grade: the book improves]
R --> G
G -->|clean| DONE[Nothing to do today]
In the demo, an at-risk book graded D becomes A− after the desk rotates a memecoin position into a tokenized blue-chip plus a cash buffer and de-risks a leveraged perp. Real residual risk is left on the board, and shown honestly.
Leverage · Biggest position · Move-together risk · Cash buffer · Spread
(diversification) · Leverage cost · Drop from peak. Each is plain-labelled, has a
one-line explanation, and is scored deterministically. See lib/vitals.ts
and the test suite in lib/vitals.test.ts (npm run test:vitals).
The deployed desk runs live and on-chain. Every step in the audit trail is tagged.
on-chain only if its transaction actually landed; a step that
cannot land is honestly marked sim, never faked.
(example tx ·
agent wallet)Running the code locally with no wallet configured falls back to a clearly labelled simulation, so anyone can try it. The deployed instance is live on-chain.
An earlier version of this idea only worked while a session was open on the builder's own laptop. When the laptop slept, the agent went dark. Second Opinion Pro is built the opposite way: the agent brain is serverless, its state lives in a serverless database, and two independent heartbeats advance work unattended. Close your laptop and it keeps running.
| Layer | What | Where |
|---|---|---|
| Diagnosis | Deterministic seven-vitals grading (A+ to D) | lib/vitals.ts |
| Portfolio | Demo books + live X Layer balance read | lib/portfolio.ts, lib/chain.ts |
| Agent economy | Plan → hire (A2A) → pay (x402) → execute → settle | lib/agents.ts |
| Orchestrator | Always-on state machine, guardrails, kill switch | lib/desk.ts |
| Chain | X Layer client, agent wallet, on-chain anchoring | lib/chain.ts |
| Storage | Redis (Upstash / Vercel KV) or in-memory fallback | lib/store.ts |
| API | Serverless routes: scan / approve / tick / kill / chain | app/api/* |
| Interface | Instrument-panel UI | components/desk-ui.tsx |
Stack: Next.js 14 on Vercel, React, TypeScript, viem, X Layer (chain 196).
npm install
npm run dev # http://localhost:3838
npm run test:vitals
Runs with zero config in honest simulation mode. See .env.example
for going always-on (Upstash) and live on-chain, and DEMO.md for the
demo script.
SUBMISSION.md — the full submission writeup.QA_BRIEF.md — how to test it, and expected-vs-bug behaviour.DEMO.md — the demo script.