Crypto API docs
The editorial ground truth for crypto agents. The Crypto Brain from Blockchain Academics is one API key over our newsroom, research desk, entity dossiers, live market and on-chain data, grants, events and jobs: 105 tools in 19 categories, served as a REST API and as an open Model Context Protocol server for Claude, ChatGPT, Cursor, LangChain, Eliza and any other agent runtime. Every answer is dated, cited and cross-referenced instead of recalled from pretraining.
Published as @blockchainacademics/mcp on npm and bca-mcp on PyPI. MIT licensed. One command to install.
Start here
@blockchainacademics/mcp into Claude Desktop in a minute. Also works in Claude Code, Cursor, Windsurf and any stdio client.What you get
- 105 typed tools across news and research, market data, on-chain, sentiment and crypto Twitter, proprietary indicators, agent-backed generation, directories, fundamentals, chains, microstructure, regulatory, security, memes, compute, services, history and corpus meta.
- 6 proprietary indicators, among them the coverage index, narrative strength, sentiment velocity, editorial premium, KOL influence and risk score, computed on our own corpus.
- 6 agent-backed tools: due diligence memos, tokenomics models, whitepaper summaries, contract translation and keyword subscriptions (saved today; the watcher that sends their webhooks is not built yet). Long-running jobs return a job id, and
get_agent_jobreturns the result when it is done. - Attribution envelope on every response: every call returns
{data, attribution: {citations: [...]}, meta}. Each citation carriescite_url,as_ofandsource_hash. Your agent never has to guess where a fact came from, and your users can always click through. See Response envelope.
Why this over pretraining?
Pretraining is a frozen snapshot. Blockchain Academics is live editorial: new articles every day, 3,800+ archived pieces and reports, and an entity graph of crypto projects, tokens, people and institutions, all reconciled to a single source of truth, all timestamped, all citable.
When an agent calls search_news("circle IPO") it gets dated titles and summaries with each article as its citation: a cite_url that opens the article, with a ?src= tag on our own pages so the agent's readership is attributed correctly. No hallucinated dates, no made-up quotes, no stale pretraining.
Supported clients
| Client | Transport | Status |
|---|---|---|
| Claude Desktop | stdio | Supported: quickstart |
| Claude Code | stdio | Supported: one command |
| Cursor / Windsurf | stdio | Supported: config file |
| LangChain (Python) | stdio subprocess | Supported: quickstart |
bca CLI (PyPI) | HTTP | Supported: CLI |
| Eliza | MCP plugin | Supported through @elizaos/plugin-mcp; dedicated plugin in development |
| Virtuals Protocol | HTTP | On the roadmap; call the REST API meanwhile |
| Custom | stdio or REST | Supported: MCP quickstart, REST API |
Pricing & attribution
Free and Starter include the core read tools, rate-limited. Pro unlocks the proprietary indicators and most agent-backed tools, and Team the rest. Plans and quotas are on Pricing; the behavior that changes between tiers is on Tiers & quotas.
Attribution is required on Free and Starter. See Attribution & licensing for the citation contract and the
llms.txtreference.
Need help?
- Questions: GitHub Discussions, public Q&A where every answer stays searchable
- Bugs: GitHub issues on
blockchainacademics/bca-mcp, alongside pull requests and the roadmap - Account and billing: [email protected]
- Paid plans: email the same address about anything, answered within the response time your plan lists on Pricing
- Status: blockchainacademics.com/crypto-api/status
Discussions and issues are public: post your key prefix, never the full key.