Reference
Frequently Asked Questions
Hive supports remote MCP, REST, CLI, and adapter access through one API key. The questions below cover setup, pricing, authentication, discovery, rate limits, and the live tool catalog.
Get a Hive API key → See pricing
Is Hive Intelligence free?
Yes. Hive has a Free Demo tier: 10,000 monthly credits, 30 requests per minute, 5 API keys, access to every tool across every category. No credit card required. Paid tiers (Analyst $129/mo, Pro $499/mo, Enterprise custom) increase the credit budget, raise the rate limit, and add support. Every plan keeps full tool access.
What data does Hive return?
Hive returns live crypto market and on-chain data through one execution contract: spot and OHLC prices, derivatives and funding rates, DeFi TVL and yields, wallet balances and transaction history (EVM and Solana), token metadata and security checks, DEX pool and trade data, NFT collection stats (including Solana compressed NFTs), Solana RPC primitives (slots, priority fees), and Polymarket prediction market data. Hive execution metadata includes fetched_at so agents can reason about when the tool execution retrieved data.
Who uses Hive Intelligence?
AI agent builders: research agents that monitor crypto markets, trading bots that need live prices plus DEX liquidity, wallet analyzers that work across chains, security scanners that call pre-signing checks, and DeFi yield comparators. Typical callers are Claude Desktop, Claude Code, Cursor, OpenAI Responses API, Windsurf, VS Code (GitHub Copilot Chat), Codex CLI, and Gemini CLI agents. Hive is built for agent tool calls, runtime discovery, and programmatic workflows.
How many tools does Hive have?
Hive exposes 375 callable tools in the current catalog: 357 provider tools across 10 category-scoped MCP endpoints plus 18 Hive-native stateful monitoring, memory, alert, report, and B2B subject tools. Because new tools and providers ship continuously, the canonical count lives at GET /api/v1/tools or on Live Catalog. At the root MCP endpoint, a compact 13-tool discovery layer surfaces search_tools, 10 category listings, get_api_endpoint_schema, and invoke_api_endpoint.
Does Hive work with Claude Code?
Yes. Run claude mcp add --transport http hive https://mcp.hiveintelligence.xyz/mcp --header "Authorization: Bearer YOUR_KEY" and Hive tools show up in your next session. Use --scope user to install globally or --scope project to commit to the repo's .mcp.json. See Install in Claude Code for the full flow.
What is the difference between the MCP and REST API?
Same server, two surfaces. MCP (Model Context Protocol) is the streaming JSON-RPC interface AI agents speak natively. The root MCP endpoint keeps tools/list compact for discovery and routing, while category endpoints expose direct scoped tools. REST is a plain HTTP surface (POST /api/v1/execute, GET /api/v1/tools) for backends and scripts that do not implement MCP. Both hit the same live catalog and both consume one credit per tool call.
What is a managed MCP server?
An MCP server is a backend that exposes tools over the Model Context Protocol for AI agents to discover and call. "Managed" means Hive runs the server, authenticates the upstream providers, handles rate limiting and retries, normalizes response shapes, and ships new tools as we add providers. You never touch the provider APIs directly. Self-hosting would mean running the server, rotating provider keys, and maintaining the schema yourself.
How does Hive handle rate limits?
Hive enforces two layers. First, each Hive API key has a per-minute request limit and your account has a monthly credit budget shared across all your keys, based on your plan (Demo: 30 req/min, 10,000 credits/month). Hitting the per-minute limit returns HTTP 429 RATE_LIMITED with a Retry-After: 60 header; exhausting monthly credits returns HTTP 429 QUOTA_EXCEEDED. Second, Hive internally respects each upstream provider's rate limit; upstream failures surface as HTTP 503 with code PROVIDER_UNAVAILABLE. Both layers are documented in Rate Limits and Errors.
What does the live server expose?
The current Hive server exposes:
- 10 category-scoped MCP endpoints
- 357 categorized provider tools
- 18 Hive-native stateful tools
- 375 callable tools in the full live catalog
Use Tools Reference for scoped endpoints and Live Catalog for the full callable catalog.
Why do the root endpoint and category endpoints look different?
The root /mcp endpoint is designed for discovery-heavy agent workflows. Category endpoints expose smaller direct tools/list results. The full wrapped tool catalog is discoverable through:
GET /api/v1/toolshive://toolsinvoke_api_endpointfrom root/mcp- direct
tools/callon category endpoints
Is the documentation static or live?
The public docs are curated, but the product itself is backed by the current live Hive catalog. That means the dashboard, CLI, and REST surfaces are intended to reflect real executable tools and not just static marketing examples.
How do I authenticate?
Send Authorization: Bearer YOUR_HIVE_API_KEY. The legacy x-api-key: YOUR_HIVE_API_KEY header is still accepted for backward compatibility but is no longer recommended for new integrations.
What capabilities are live today?
- Market data (prices, caps, OHLC, trending)
- DeFi analytics (TVL, yields, fees, bridges, stablecoins)
- Security & risk (token, wallet, dApp, phishing, rug-pull, simulation, LP locks)
- On-chain DEX analytics (pools, pairs, trade flow)
- Exchange data (spot, perps, funding rates)
- Portfolio & wallet analytics (balances, transactions, PnL, approvals)
- NFT analytics (collections, floors, sales, trades)
- Solana depth (native RPC, Digital Asset Standard, compressed NFTs, parsed transactions)
- Prediction markets (Polymarket events, trader positions)
- Network & infrastructure (blocks, logs, gas, chain status)
See Data Coverage for full scope.
What is the current REST payload shape?
Use:
{
"tool": "get_price",
"args": {
"ids": "bitcoin",
"vs_currencies": "usd"
}
}The older toolName / arguments shape is legacy documentation and is no longer the preferred format.