Skip to main content
Glama

degenscan-intel

Cross-asset event intelligence for autonomous agents. One normalized feed of everything that moves prices — natural events, regulators, central banks, federal rules, corporate filings, trading halts, on-chain shocks, prediction-market shifts — scored against an exposure graph into per-asset impacts, and sold per call in USDC via x402 or by API key.

Agents don't want a 3D globe. They want events_since("4h", universe=["NVDA","BTC","CL"]) for half a cent.

Spec and rationale: docs/SPEC.md.

Quick start

npm install                     # Node ≥ 22.13 required (node:sqlite)
cp .env.example .env            # optional; runs keyless in free mode by default
npm run probe                   # hit every connector once, dry-run, report reachability + sample scored events
npm run dev                     # ingest scheduler + MCP (POST /mcp) + REST (/v1) on :8787
curl 'localhost:8787/v1/events?since=4h&universe=NVDA,TSM,BTC&min_confidence=0.2'
curl  localhost:8787/v1/impact/NVDA?since=24h
curl  localhost:8787/v1/graph/MSTR
curl  localhost:8787/v1/regime
curl  localhost:8787/v1/sources

MCP (streamable HTTP, stateless): POST http://host:8787/mcp. Stdio for local agents: npm run mcp:stdio.

Related MCP server: orchard-data-mcp

What's inside

Layer

File(s)

Notes

Schema

src/schema.ts

Event, Impact, Asset, Universe (zod) — the contract agents consume

Universe

src/universe/

~110 curated equities (tags: sector/hq/rev/input/reg/theme) + ETFs, indices, 15 crypto, commodities, FX, rates. Daily top-100-by-volume refresh (Nasdaq screener → Yahoo fallback → seed)

Exposure graph

src/graph/

~230 curated edges + tag-derived edges + 30 critical facilities with lat/lng (fabs, ports, straits, refineries). propagate() = weighted BFS depth 3

Engine

src/engine/

entity linking (tickers, names, countries, regulators, commodities, themes, geo→facility), deterministic direction inference, session calendar (tradable_now, next_open, US holidays)

Connectors

src/ingest/connectors/

40 keyless sources: USGS, GDACS, EONET, NHC, NWS, SWPC · SEC EFTS 8-K by item, EDGAR Atom (Form 4/13D/S-1), SEC/Fed/ECB/BoE/BoJ press, Fed speeches, Federal Register API, FTC, DOJ, FDA, CFTC, FCC, White House, EU Commission, CISA KEV, OFAC, TreasuryDirect · Nasdaq halts, Nasdaq earnings, PR Newswire, Business Wire, GlobeNewswire, Google News per ticker, NHTSA · DefiLlama hacks + stablecoin supply, CoinGecko abnormal moves, Binance announcements, Coinbase status, mempool fees, Polymarket, GDELT DOC

Store

src/store/db.ts

SQLite via node:sqlite (built into Node ≥ 22.13, zero native builds) + FTS5; cross-source corroboration by title fingerprint; per-asset impact index; source health; billing log

Server

src/server/

Fastify REST + MCP; x402.ts payment gate (402 → verify → settle via facilitator), API keys, free daily quota; pricing.ts

Tools & pricing

Tool

Price

What it answers

events_since

$0.005

what happened since t that touches my universe (past t = backtest)

impact_for

$0.003

net directional pressure on one asset + source events

exposure_graph

$0.002

who/what an asset is exposed to

regime_snapshot

$0.01

venues open, 24h pressure ranking, high-severity events, prediction markets

explain

$0.02

rationale for one event

pulse

$0.001

cheapest probe: last-hour event counts by class, top-3 severe, venues open

price_for

$0.001

price probe, no key: Hyperliquid mark/mid/oracle + Coinbase spot, 24h change, basis

funding_alerts

$0.001

perps with extreme funding now + predicted per venue

whale_moves

$0.002

large USDC/USDT transfers on Base/Ethereum with exchange labels and flow tags

polymarket_top

$0.002

most active Polymarket markets: odds, 24h change, volume, liquidity

derivs_for

$0.003

perp microstructure for one coin (Hyperliquid public API): funding, OI, premium, predicted funding by venue, flags + event pressure

news_for

$0.002

headlines on one asset with tier, corroboration, heuristic sentiment

filings_for

$0.002

SEC EDGAR filings on one issuer (8-K, Form 4, 13D/G, S-1)

calendar

$0.002

upcoming macro prints, FOMC, auctions, earnings (N days)

brief

$0.10

premium one-call pre-trade briefing per asset (replaces six calls)

oracle_board

$0.002

daily board of calibrated probabilities for standing questions (BTC/ETH/SOL targets, FOMC, top Polymarket) — cached, no waiting

oracle_forecast

$0.25

calibrated YES-probability for any binary question: Monte Carlo of LLM agent societies + base-rate-anchored expert panel on live Intel data; async (202 → poll free); 80% interval, edge vs market, drivers, commitment hash. Public Brier record at /v1/oracle/track-record (free)

polymarket_context

$0.01

evidence pack for one prediction market: current odds + primary-source events in the feed that bear on the question

universe, sources_status

free

discovery & transparency

Free trial: header X-Free-Trial: 1 → 100 calls/day/IP without payment (MCP tools/call gets it automatically). Otherwise priced routes answer 402 with x402 requirements, or accept X-API-KEY.

Networks: USDC on Base (eip155:8453) and, when configured, Solana mainnet — both offered in the same 402, gasless via PayAI.

Three ways to pay

Who

How

Endpoint

Autonomous agent, per call

x402 v2 — answer the 402 with USDC on Base; any x402 client does it automatically

any priced endpoint

Autonomous agent, prepaid key (no human, no card)

one USDC payment → { api_key } with a lifetime budget: pack_1k $5 / 1,000 calls · pack_10k $40 / 10,000 · pack_100k $300 / 100,000

POST /v1/keys/x402/<pack> → then header X-API-KEY; balance at GET /v1/keys/me

Human operator, card

Stripe subscription Starter $29 (10k/mo) · Pro $199 (200k/mo) → key shown once

GET /v1/plans → checkout → /v1/keys/claim

For agents & indexers: /.well-known/x402 (resources, prices, payTo) · /openapi.json · /llms.txt · /wallets.json (our own wallets, excluded from metrics). Install the skill: npx skills add tradewr333-lgtm/degenscan-intel · SDKs: npm i @degenscan/intel (packages/js) · pip install degenscan-intel (packages/py).

Agent skill: skills/degenscan-intel/SKILL.md (also served at /skill.md) — when to call which tool, the recommended loop, how to read confidence, how to pay. Drop it into a Claude/Cursor skills folder or point the agent at the URL.

Public usage metrics: /v1/metrics (JSON) · /v1/metrics.csv — one row per week since 2026-09-27: free/API-key/paid calls, unique paying wallets, USDC revenue with on-chain tx hashes, active Stripe subscriptions. Owner and test wallets are listed explicitly and excluded from customers and revenue.

Configuration

See .env.example. Nothing is required to run. For paid mode set X402_PAY_TO (Base address) and, for Base mainnet, a facilitator that supports it (Coinbase CDP or self-hosted). INTEL_UA must include a contact e-mail — SEC blocks anonymous clients.

Deploy (Render)

render.yaml defines one web service (server + scheduler in-process, persistent disk for SQLite). Dockerfile for anywhere else.

Design principles

  1. Primary sources first. SEC, Fed, USGS, Nasdaq before any news site. Media is corroboration (corroboration.count), never origin.

  2. Deterministic hot path. No LLM between event and impact — latency < 200 ms, cost ≈ 0. explain is the only place a model may be used.

  3. Everything auditable. raw_ref points at the source document; path shows the graph route; sources_status is public.

  4. Universe is versioned. Every response carries universe_version so backtests are reproducible.

  5. Legal by construction. Only public/government/issuer data; no real-time equity quote redistribution; unofficial APIs (Nasdaq, Yahoo, Reddit) isolated as best-effort connectors.

Roadmap

See SPEC §8. Next: deploy + live probe → real x402 settlement ✅ → registries (MCP Registry, Smithery, Glama, x402 Bazaar) ✅ → 10-K-driven graph expansion → webhooks → Degenscan bot integration.

Operator, disclaimer & data licensing

Operated by Marbella Collins LLC (Florida, USA) · contact@degenscan.io. Built and operated by Renato Pereira Chagas, who specified the product, made the design and commercial decisions and operates the service; the code was written with AI assistance (Claude).

Not investment advice. Degenscan Intel is an information and analytics service. Impact scores are deterministic heuristics over public events; they are not recommendations to buy or sell any asset and carry no guarantee of accuracy or timeliness. You are solely responsible for your trading decisions.

Data licensing. All connectors read public, keyless sources; the service redistributes normalized event metadata (titles, timestamps, links) and its own derived scores, never bulk source content or real-time quotes. Government and issuer sources (SEC, Fed, Federal Register, USGS, NOAA, agency press) are public domain or freely redistributable. Sources marked best-effort in sources_status use unofficial or terms-limited endpoints (Nasdaq halts/earnings pages, Yahoo screener fallback, CoinGecko public API) and are subject to those providers' terms; they are isolated so the service degrades gracefully if any is removed.

Credits

Event-source ideas and the supplier-overlay concept trace to OSIRIS by simplifaisoul (MIT). This project shares no code with it and no UI; it is the data/agent layer OSIRIS never had.

License: MIT — © 2026 Marbella Collins LLC / degenscan.io

Related MCP Connectors

Related MCP Servers

  • F
    license
    Not graded
    quality
    D
    maintenance
    56 pay-per-call MCP endpoints for AI agents. Market signals, macro economics, crypto/DeFi, geopolitical intelligence, SEC filings, GitHub velocity, sanctions screening. USDC on Base Mainnet via x402.
    -
  • A
    license
    A
    quality
    C
    maintenance
    Pay-per-call MCP server for market data with USDC micro-payments, featuring unique ML market-regime forecasts, news catalyst radar, and a live-verified algorithmic trading track record.
    12
    36 npm
    MIT
  • F
    license
    Not graded
    quality
    B
    maintenance
    MCP server exposing 1,000+ pay-per-call API endpoints across agent infrastructure (memory, coordination, secrets, verification), data, compute, finance, weather, geography, and reference categories — payments via x402 protocol in USDC on Base.
    -
  • A
    license
    A
    quality
    A
    maintenance
    Pay-per-call ($0.005–$0.03 USDC) market, on-chain, and prediction-market data API for AI agents and trading bots via the x402 protocol — no signup, no API key. Exposed as a remote MCP server with 15 tools: pre-trade token security (honeypot/liquidity checks), kimchi premium, funding rate APR, DEX slippage, Polymarket arbitrage & liquidity audits, and Hyperliquid HIP-4 prediction-market odds.
    4
    15
    MIT