mcpField
D
Degenscan Intel

Degenscan Intel

Remote
degenscan-intel.onrender.comFinance0Updated 2d ago

About Degenscan Intel

Degenscan Intel provides normalized cross-asset event and market data for autonomous trading agents, linking public-source events to asset impacts through an exposure graph. It offers MCP and REST access with per-call USDC or API-key billing.

Key features

  1. 01Cross-asset event feed
  2. 02Per-asset impact scores
  3. 03Exposure graph traversal
  4. 04Market and derivatives tools
  5. 05Public-source event corroboration

Use cases

  1. 01Monitor events across a trading universe
  2. 02Assess event pressure on one asset
  3. 03Review exposures and event paths
  4. 04Fetch market and derivatives indicators

MCP support

Exposes tools through a stateless Streamable HTTP MCP endpoint, with stdio available for local agents. Tools cover event queries, asset impacts, exposure graphs, market data and related analysis.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://degenscan-intel.onrender.com/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://degenscan-intel.onrender.com/mcpAuth required

Tools

35
ParameterTypeRequiredDescription
qstringNoFull-text query over title/summary (FTS5 syntax), e.g. "tariff OR sanction".
kindsarrayNoEvent kinds or prefixes: ["reg.", "corp.8k", "nat.quake"].
limitintegerNoMax events returned (1..200).
sincestringNoWindow start: "30m", "4h", "2d" or ISO-8601. Past values work identically (backtesting).
untilstringNoWindow end (ISO-8601). Default now.
universearrayNoAsset ids to filter impacts by, e.g. ["NVDA","BTC","CL"]. Omit for all.
min_severitynumberNoDrop events below this severity (0..1). 0.5 keeps market-moving events only.
min_confidencenumberNoMin impact confidence (0..1) for the universe filter. 0.4 is a sensible threshold for acting.

Details

From the project README

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.

What's inside

LayerFile(s)Notes
Schemasrc/schema.tsEvent, Impact, Asset, Universe (zod) — the contract agents consume
Universesrc/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 graphsrc/graph/~230 curated edges + tag-derived edges + 30 critical facilities with lat/lng (fabs, ports, straits, refineries). propagate() = weighted BFS depth 3
Enginesrc/engine/entity linking (tickers, names, countries, regulators, commodities, themes, geo→facility), deterministic direction inference, session calendar (tradable_now, next_open, US holidays)
Connectorssrc/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
Storesrc/store/db.tsSQLite 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
Serversrc/server/Fastify REST + MCP; x402.ts payment gate (402 → verify → settle via facilitator), API keys, free daily quota; pricing.ts

Tools & pricing

ToolPriceWhat it answers
events_since$0.005what happened since t that touches my universe (past t = backtest)
impact_for$0.003net directional pressure on one asset + source events
exposure_graph$0.002who/what an asset is exposed to
regime_snapshot$0.01venues open, 24h pressure ranking, high-severity events, prediction markets
explain$0.02rationale for one event
pulse$0.001cheapest probe: last-hour event counts by class, top-3 severe, venues open
price_for$0.001price probe, no key: Hyperliquid mark/mid/oracle + Coinbase spot, 24h change, basis
funding_alerts$0.001perps with extreme funding now + predicted per venue
whale_moves$0.002large USDC/USDT transfers on Base/Ethereum with exchange labels and flow tags
polymarket_top$0.002most active Polymarket markets: odds, 24h change, volume, liquidity
derivs_for$0.003perp microstructure for one coin (Hyperliquid public API): funding, OI, premium, predicted funding by venue, flags + event pressure
news_for$0.002headlines on one asset with tier, corroboration, heuristic sentiment
filings_for$0.002SEC EDGAR filings on one issuer (8-K, Form 4, 13D/G, S-1)
calendar$0.002upcoming macro prints, FOMC, auctions, earnings (N days)
brief$0.10premium one-call pre-trade briefing per asset (replaces six calls)
oracle_board$0.002daily board of calibrated probabilities for standing questions (BTC/ETH/SOL targets, FOMC, top Polymarket) — cached, no waiting
oracle_forecast$0.25calibrated 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$0.01evidence pack for one prediction market: current odds + primary-source events in the feed that bear on the question
universe, sources_statusfreediscovery & 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

WhoHowEndpoint
Autonomous agent, per callx402 v2 — answer the 402 with USDC on Base; any x402 client does it automaticallyany 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 / 100,000POST /v1/keys/x402/<pack> → then header X-API-KEY; balance at GET /v1/keys/me
Human operator, cardStripe subscription Starter $29 (10k/mo) · Pro $199 (200k/mo) → key shown onceGET /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) · [email protected]. 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

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.