Agent-first market intelligence

One brain for your trades and your agents.

Ask any market — Crypto, US equities & options, India equities. MarketIntell returns a sourced, sized, stress-tested setup with calibrated confidence — then proves itself with a verifiable on-chain record. Delivered to you or your agents over chat, REST, and MCP.

Track record: +$216 on $2,731 (+7.9%) · 8/8 resolved · audit the full record ↗

Small sample · real USDC · every position on-chain · wins and losses.

Free signup, no card.

Under the hood27+live data services17MCP tools100%of claims cited4agent surfacesScoredon close — wins & losses
External services27+Specialist agents26Polymarket bot+$216 on $2,731 (+7.9%) · 8/8 resolved · on-chainMedian response~20sSource citationsEvery claimLive since2026-03-16External services27+Specialist agents26Polymarket bot+$216 on $2,731 (+7.9%) · 8/8 resolved · on-chainMedian response~20sSource citationsEvery claimLive since2026-03-16External services27+Specialist agents26Polymarket bot+$216 on $2,731 (+7.9%) · 8/8 resolved · on-chainMedian response~20sSource citationsEvery claimLive since2026-03-16

The problem

Tired of 12 tabs and a hunch?

A terminal floods you with data but never tells you what it means. A generic chatbot sounds confident, then invents the price levels. So you stitch together charts, funding, flows and headlines yourself — every single time. There's a faster way to get to a decision.

One brain, two surfaces

Built for humans and agents.

The same reasoning engine, the same live data, the same citations. Only the surface changes: a rich card for a person, structured JSON for a machine.

For humans

Decision-ready cards

Ask in plain language and get a card you can act on — never a wall of text. Card types span the whole trading loop, each sourced and scored.

Trade setupLiquidation heatmapSmart-money flowPortfolio riskPosition P&LConfidence breakdownand more across the loop
For agents

The same brain, as a tool

Drop MarketIntell into any agent stack. It speaks the protocols your tools already use — and returns clean JSON with a citation behind every number.

skill.mdMCP serverCLIREST + SSE APISigned webhooksWebSocket stream

What it covers

The whole loop, not one screen.

Every answer is reasoned through live data, scored for confidence, and backed by citations you can drill into. No vibes, no hallucinated price levels.

01
Discover
Scan, relative strength, watchlist, whale & unusual activity
02
Analyze
Technicals, structure, derivs, on-chain, fundamentals
03
Decide
Sourced setup, confidence, sizing, confluence
04
Execute
Order ticket, bracket / OCO, scale-in ladder
05
Manage
Live P&L, alerts, trailing stops, scale-outs
06
Risk
Portfolio, exposure, correlation, margin health
07
Review
Journal, performance analytics, post-mortem

Sourced trade setups

Direction, entry, stop, and targets with realized R:R — assembled from 27+ live data services, never from memory.

Calibrated confidence

Every signal carries a confidence score that's audited against realized outcomes — predicted and actual track within points.

Autopilot agents

Spin up bots that watch the market on a schedule or trigger, chat with the brain, and place trades through your broker.

Auditable track record

Every signal scored on close. Accuracy by timeframe, by domain, and by confidence bucket — open to inspection.

Autopilot, with guardrails

Trading real money is opt-in and bounded. Custody depends on how you connect: OAuth brokers (Coinbase Advanced, Kraken, Alpaca) keep funds in your account; Hyperliquid uses an on-chain agent you approve; crypto spot on Base uses a dedicated agent wallet whose encrypted private key we hold on our servers. Execution runs through your connected paths via revocable, scoped API access. Every agent runs inside hard limits you set — max position size, daily loss cap, allowed symbols — with a kill-switch and an approval mode that asks before it places anything. Leave trade execution off and Autopilot stays advisory: it surfaces setups; you place the orders yourself.

How it works

Question in. Sourced answer out.

Whether the caller is a person or an agent, the pipeline is identical — and it shows every source.

1

Ask anything

A ticker, a thesis, a “what if” — in the app, or as an MCP / CLI / API call from your agent.

› Is NVDA's skew bullish into GTC?
2

The brain reasons

It fans out across 27+ live data services, cross-checks the signals, and scores its own confidence.

✓ price · derivs · flow · on-chain · news
3

Get a sourced result

A rich card for a human; cited JSON for an agent. Same answer, same evidence trail.

› Long · 3,438 · stop 3,265 · 82%

Who it's for

One brain, three jobs.

Active traders, funds, and the engineers building the next generation of trading agents — same engine, different surface.

Active traders

Self-directed · crypto + equities

  • Find and stress-test setups fast
  • Size positions to a fixed risk
  • Track every call on an honest scoreboard

Funds & desks

Teams · shared workflow

  • Consistent, sourced analysis across the team
  • Auditable confidence & calibration
  • SSO & single-tenant on Enterprise

Agent builders

Developers · automation

  • Native MCP / CLI / REST + SSE surfaces
  • Cited JSON for trustworthy automation
  • Signed webhooks & high rate limits