The MCP every crypto agent needs.

106 tools across CoinGecko, DefiLlama, Etherscan, Helius, DexScreener and our own corpus — one auth, one invoice, one citation.

npx @blockchainacademics/mcp

Get a free key Browse 106 tools Read docs

Free tier: 2,000 calls/mo. No card.

claude-desktop · bca-mcp

❯ agent research ethereum restaking landscape

› calling bca.get_mcp_recipe({ mission: 'narrative_research' })
› calling bca.search_news({ q: 'restaking', limit: 10 })
› calling bca.get_entity({ slug: 'eigenlayer' })
› calling bca.get_coverage_index({ entity: 'eigenlayer', window: '30d' })
› calling bca.generate_research_report({ topic: 'restaking', depth: 'deep' })

✓ 5 tool calls · 1.28s · cited 12 BCA articles
  cite_url: https://blockchainacademics.com/articles/…

Built in the open. Shipping every week.

npm@blockchainacademics/mcp v0.1
GitHubOpen source
Anthropic MCPPending reviewdirectory
Corpus3,501 · 200 · 43articles · entities · lessons
Uptime99.9% target

The problem

Every crypto agent reinvents the same plumbing.

Fragmented sources.

You wire CoinGecko, DefiLlama, Etherscan, Helius, DexScreener separately. Five keys, five schemas, five rate limits.

No editorial layer.

Raw feeds don't tell you what matters. Agents drown in data and miss the narrative.

No attribution.

Your AI cites Wikipedia. Meanwhile crypto-native sources never make it into the answer.

Architecture

One auth boundary. Everything behind it.

We aggregate what's free. We build what isn't. You call one endpoint.

Aggregated

CoinGecko
DefiLlama
Etherscan
Helius
DexScreener
Fear & Greed
1inch
Polymarket

Public + free-tier sources

Proprietary

3,501 articles
200 entities
43 lessons
6 composites
11 in-house agents

Our moat — not proxied

Surfaced as

106 MCP tools
REST API
stdio + SSE

One schema, one auth

Over: your API key

Tool catalog

106 tools. Nine categories. One schema.

Every response is wrapped: {data, cite_url, as_of, source_hash}. Time-travel is a query param.

The moat. 3,501 articles, ~200 entities, 9 courses, 43 lessons — all FTS-indexed and time-travelable.

search_newsfree
Full-text search across the 3,501-article corpus.
Live
get_articlefree
Full article body + citations + entity graph.
Live
get_entityfree
Dossier for any chain, project, person, or ticker.
Live
list_entity_mentionsfree
Mention timeline + sentiment per entity.
Live

See all 7 content tools

RequestTypeScript · Python

await mcp.call("search_news", {
  q: "EigenLayer restaking",
  limit: 3,
  as_of: "2026-04-01"
})

Responsecite_url · as_of · source_hash

{
  "data": [
    {
      "slug": "eigenlayer-restaking-primer",
      "title": "EigenLayer Restaking: The AVS Landscape",
      "summary": "A structural map of the restaking stack…",
      "entities": ["eigenlayer", "ethereum", "avs"],
      "sentiment": 0.42,
      "published_at": "2026-03-28T09:12:00Z"
    }
  ],
  "cite_url": "https://blockchainacademics.com/articles/eigenlayer-restaking-primer?src=claude",
  "as_of": "2026-04-01T00:00:00Z",
  "source_hash": "sha256:8f3a…c21b"
}

Browse all 106 tools

Proprietary indicators

Six composites the rest of the market can't ship.

Built from our editorial corpus and entity graph. Not feeds we proxy — composites we own.

BCA Coverage Index

Formula

Σ(mentions × source_weight × recency_decay)

Pre-price coverage surge. Surfaces the accumulation window before the crowd notices.

Narrative Strength

Formula

Eigenvector centrality in co-mention graph

Which narratives are consolidating vs fading. Rolling 7 and 30-day windows.

Sentiment Velocity

Formula

d(sentiment_bucket)/dt, smoothed

Early reversal detection. Signals momentum shifts before price moves.

Editorial Premium

Formula

corr(price_return, coverage_delta) lagged −1 to +3d

Measures whether smart money reads BCA. The self-referential alpha metric.

KOL Influence

Formula

reach × engagement × historical_accuracy

Surfaces actually-predictive KOLs — not just the loud ones. Built from our 88-contact CRM.

Risk Score

Formula

w₁·regulatory + w₂·liquidity + w₃·team + w₄·audit

Single-number risk. Same schema as investment memos, for any entity in the corpus.

Raw component values available on Enterprise. Custom weightings available on Team.

Agent workflows

Agents don't want tools. They want recipes.

Call get_mcp_recipe(mission) — get a prescribed DAG of tool calls back. The first MCP that teaches itself.

Due diligence a token

Tokenomics, audits, team, risk, narrative — one call.

const plan = await mcp.call("get_mcp_recipe", {
  mission: "due_diligence",
  entity: "berachain"
})
// → [get_tokenomics, get_audits, get_team, get_risk_score, get_coverage_index]

Research a narrative

Coverage + sentiment + entity graph across 30 days.

const plan = await mcp.call("get_mcp_recipe", {
  mission: "narrative_research",
  topic: "restaking",
  window: "30d"
})
// → [search_news, list_entity_mentions, get_narrative_strength, get_sentiment]

Prep a TGE

KOL picks + airdrop calendar + marketing template.

const plan = await mcp.call("get_mcp_recipe", {
  mission: "tge_prep",
  entity: "monad"
})
// → [get_kol_influence, get_airdrops, generate_kol_shortlist]

Audit a contract

Bytecode scan + exploit history + sanctions check.

const plan = await mcp.call("get_mcp_recipe", {
  mission: "contract_audit",
  address: "0xabc…"
})
// → [get_audits, get_exploit_history, get_sanctions_check, get_tx]

Find alpha

Sentiment velocity + editorial premium crossovers.

const plan = await mcp.call("get_mcp_recipe", {
  mission: "find_alpha",
  universe: "top_100"
})
// → [get_sentiment_velocity, get_editorial_premium, get_coverage_index]

Onboard a chain

RPCs + faucets + grants + hackathons + top projects.

const plan = await mcp.call("get_mcp_recipe", {
  mission: "chain_onboard",
  chain: "base"
})
// → [list_rpcs, list_faucets, get_grants, get_hackathons, get_ecosystem_map]

Onboarding

Install to first call in 60 seconds.

Three steps. One persistent terminal. Step three returns a real production response.

Install the MCP server

~ · bash

Runs locally. No code changes to your agent — just config.

Get a key

~ · bash

Free tier: 2,000 calls/mo. No card.

Hit production

~ · bash

$ curl https://api.blockchainacademics.com/v1/market/price?ticker=ETH \
$      -H "X-API-Key: $BCA_KEY"

200 OK · application/json

{
  "data": {
    "ticker": "ETH",
    "price_usd": 3287.14,
    "change_24h_pct": 2.31,
    "change_7d_pct": -4.18,
    "change_30d_pct": 11.77,
    "mcap_usd": 396100000000,
    "volume_24h_usd": 18420000000
  },
  "cite_url": "https://blockchainacademics.com/entities/ethereum?src=claude",
  "as_of": "2026-04-20T14:00:00Z",
  "source_hash": "sha256:2d1e4f…7fa4"
}

You're now hitting the same API used by the agents in the Anthropic directory.

Client setup

Drop-in support for every major agent runtime.

stdio for Claude Desktop + Cursor. Python SDK for LangChain. Plugins for Eliza + Virtuals. Plain HTTP / SSE for anything else.

npx @blockchainacademics/mcp

claude_desktop_config.json

{
  "mcpServers": {
    "bca": {
      "command": "npx",
      "args": ["-y", "@blockchainacademics/mcp"],
      "env": { "BCA_API_KEY": "your_key_here" }
    }
  }
}

Full docs →·GitHub →·npm →·PyPI →

Pricing

Priced like a dev tool. Built like an institutional terminal.

We absorb aggregator costs at every tier. Free means free. Pro means you get everything.

Free

free

Hobbyists & trials

Calls: 2,000 / mo
Req/sec: 1 req/sec
Agent jobs: —
Indicators: —
History: 30 days

Get a free key

Starter

starter

$9/mo

Indie builders

vs CoinGecko Analyst $129 · 93% less

Calls: 25,000 / mo
Req/sec: 5 req/sec
Agent jobs: —
Indicators: Risk Score only
History: 90 days

Start Starter

Pro

pro

$49/mo

Agent devs & prop desks

vs CoinGecko Lite $499 · 90% less

Calls: 250,000 / mo
Req/sec: 30 req/sec
Agent jobs: 10 / day
Indicators: All 6
History: 1 year

Go Pro

Team

team

$199/mo

Funds, SaaS, data teams

vs CoinGecko Pro $999 · 80% less

Calls: 1,500,000 / mo
Req/sec: 100 req/sec
Agent jobs: 50 / day
Indicators: All 6 + custom weights
History: 3 years

Go Team

Enterprise

enterprise

Custom

Funds, banks, exchanges

Under Messari Enterprise · typically 80% less

Calls: Custom
Req/sec: Negotiated
Agent jobs: Unlimited
Indicators: All 6 + raw components
History: Full archive

Talk to us

See the full matrix

Comparison

The honest comparison.

We're not for everyone. If you need institutional research reports, buy Messari. If you need one crypto agent API, read this row.

Feature	BCA Crypto Brain	CoinGecko	Messari	Kaito	DEXTools
Price entry tier	$9/mo	$129/mo	$500/mo	~$1k/mo	$70/mo
MCP-native	Yes	No	No	No	No
Proprietary corpus	3,501 articles	No	Yes	Partial	No
Agent-backed tools	11 skills	No	No	No	No
On-chain aggregation	Etherscan+Helius	Limited	Limited	No	DEX only
DEX pair data	DexScreener	Partial	No	No	Yes
Proprietary indicators	6 composites	No	2	1	0
Time-travel (as_of)	Yes	No	Partial	No	No
Citation envelope	Every response	No	No	No	No
Attribution for AI	llms.txt + cite_url	No	No	No	No
Free tier	2,000 calls/mo	30 calls/min	None	Trial	Limited
Single auth	Yes	N/A	N/A	N/A	N/A

Start free

FAQ

Questions that actually get asked.

MCP (Model Context Protocol) is how AI agents call external tools. One server exposes tools any MCP-compatible agent can use. We're the one for crypto.

Put BCA in your agent's context window.

Grab a key. You'll hit the API inside 60 seconds.

No spam. Unsubscribe from the email we send you. Rate-limited endpoint.