MINARA

Getting Started

What Minara Agent is and which path is right for you

👥 This section is for: everyone — end users picking it up for the first time, developers deciding how to integrate, and contributors setting up a dev environment. Three paths below so you can jump to yours.

Minara Agent is an AI assistant you chat with in plain language to analyze, trade, and manage money across both crypto and traditional financial markets. Ask "is BTC overbought right now?" or "buy $100 of SOL on Solana" or "show my portfolio across all chains" — the agent handles the mechanics, previews anything involving real funds, and waits for your confirmation.

📘 New here? Skim the Features overview to see what the agent can do before picking an install path. Hit an acronym you don't know? The Glossary is one click away.

Pick your path

Try it out. You're a trader, investor, or just curious. Zero coding required — skip straight to a safe sandboxed walkthrough in Your first trade.

Integrate or extend. You're a developer. Install the binary, run the HTTP gateway, add custom skills or MCP servers, tune env vars. Start with Installation, then Env vars and HTTP gateway.

Contribute. You're adding to the project itself. Read the Conventions first, then System design for architecture rationale.

The 5 pages in this section (in order)

#PageWhat you'll learn
1InstallationOne-liner installer, what the binary does, how to verify it worked
2Environment VariablesWhich env vars unlock which features; precedence rules (shell > .env > defaults); migration from legacy names
3Your First TradeWalk through a safe sandboxed buy step-by-step — no real money moved until you confirm twice
4DeploymentDocker, systemd, remote hosting, data directory pinning (dev + prod)
5TestingRunning the test suite as a contributor

What problem this solves

If you've ever:

  • Copy-pasted between 5 tabs (CoinGecko → TradingView → your wallet → a spreadsheet → Twitter) to decide whether to buy or sell — the agent does that in one prompt.
  • Set up a DCA bot manually and then forgot to check it for a month — the agent keeps a scheduled workflow running and reports back.
  • Lost money to a rug that a simple token-safety check would have caught — the agent runs that check before you confirm.
  • Tried to copy-trade a whale and had no idea what to follow — the agent can monitor specific wallets and surface actionable moves.

The goal: one natural-language agent, one memory, one safety layer — covering everything from an S&P 500 rebalance to a Solana meme-coin autopilot, and asking for your explicit confirmation on anything that moves real funds.

Next

On this page