degenscan-intel
Market-event intelligence for AI trading agents, paid per call in USDC (x402) or with an API key.
One normalized feed of price-moving events from ~40 primary sources — SEC EDGAR (8-K, Form 4, 13D), Federal Reserve, Federal Register, ECB/BoE/BoJ, FTC/DOJ/FDA/CFTC/FCC, USGS earthquakes, NOAA/NHC storms, Nasdaq halts, DefiLlama hacks, Polymarket odds, Hyperliquid perps — scored against an exposure graph into per-asset impacts (direction, confidence, path). Deterministic, sub-200 ms, every event links to its source document.
- Service:
https://intel.degenscan.io· MCP:POST https://intel.degenscan.io/mcp· OpenAPI:/openapi.json· Docs for LLMs:/llms.txt - Prices: $0.001–$0.02 per call (brief $0.10). No subscription needed. Free trial: 100 calls/day/IP.
- Operator: Marbella Collins LLC · MIT · Information and analytics only — not investment advice.
Install
pip install degenscan-intel # API key / free trial
pip install "degenscan-intel[x402]" # + per-call USDC payments from an agent wallet
20-line trading-agent loop
import os, time
from degenscan_intel import Intel
# Pick ONE:
intel = Intel(private_key=os.environ["AGENT_WALLET_PK"]) # pays USDC on Base per call via x402 (HTTP 402 -> sign -> 200)
# intel = Intel(api_key=os.environ["INTEL_API_KEY"]) # prepaid pack or Stripe plan
# intel = Intel() # free trial, 100 calls/day
book = ["BTC", "ETH", "NVDA", "MSTR", "CL"]
while True:
p = intel.pulse() # $0.001 — anything new in the last hour?
if p["high_severity"]:
for e in intel.events_since(since="4h", universe=book, min_confidence=0.4)["events"]: # $0.005
for imp in (i for i in e["impacts"] if i["asset_id"] in book and i["confidence"] >= 0.5):
brief = intel.brief(imp["asset_id"]) # $0.10 — pressure, headlines, filings, derivatives, catalysts
if not brief["tradable_now"]:
continue # venue closed -> wait for next_open
d = intel.derivs_for(imp["asset_id"]) if imp["asset_id"] in ("BTC", "ETH") else None # $0.003 — funding/OI/premium
print(imp["asset_id"], "LONG" if imp["direction"] > 0 else "SHORT", imp["confidence"], e["title"], d and d["flags"])
# -> your execution logic here
time.sleep(3600)
Async: from degenscan_intel import AsyncIntel — same methods, awaited.
Buy a prepaid key with USDC (no human, no card)
intel = Intel(private_key=os.environ["AGENT_WALLET_PK"])
key = intel.buy_pack("pack_1k")["api_key"] # $5 USDC -> 1,000 calls, lifetime. Also pack_10k ($40), pack_100k ($300)
cheap = Intel(api_key=key) # no per-call signatures from here on
cheap.key_status() # {"calls_used": ..., "calls_left": ...}
Methods
| Method | REST | Price | Returns |
|---|---|---|---|
pulse() |
GET /v1/pulse |
$0.001 | event counts last hour by class, high-severity count, venues open |
events_since(since, universe, min_confidence, limit) |
GET /v1/events |
$0.005 | events with per-asset impacts, tradable_now, next_open |
impact_for(asset, since) |
GET /v1/impact/{asset} |
$0.003 | net bias on one asset + driving events |
exposure_graph(asset, depth) |
GET /v1/graph/{asset} |
$0.002 | suppliers, countries, commodities, regulators, indices |
regime() |
GET /v1/regime |
$0.01 | venues open, 24h pressure by asset, top events, prediction markets |
explain(event_id) |
GET /v1/explain/{id} |
$0.02 | reasoning behind one impact |
polymarket(market) |
GET /v1/polymarket/{market} |
$0.01 | current odds + primary events that bear on the question |
news_for(ticker) |
GET /v1/news/{ticker} |
$0.002 | headlines with tier, corroboration, sentiment |
filings_for(ticker, forms=[...]) |
GET /v1/filings/{ticker} |
$0.002 | 8-K, Form 4, 13D/G, S-1 |
calendar(days, types) |
GET /v1/calendar |
$0.002 | FOMC, CPI, NFP, PCE, GDP, earnings, auctions |
derivs_for(symbol) |
GET /v1/derivs/{symbol} |
$0.003 | Hyperliquid funding (1h/8h/annualized), predicted funding by venue, OI, premium, 24h volume, flags |
brief(asset) |
GET /v1/brief/{asset} |
$0.10 | everything above for one asset in one call |
universe(), sources(), health(), plans(), packs() |
— | free | coverage, connector status, card plans, USDC packs |
Every paid response carries _billing: {tool, price_usd, method}; x402 responses also carry _payment_response (settlement receipt with tx hash).
How payment works (x402)
- Agent calls a priced route → server answers HTTP 402 with
PAYMENT-REQUIRED(USDC on Baseeip155:8453or Solana, amount, payTo). - The client (Coinbase's
x402package) signs an EIP-3009 USDC transfer with your wallet — gas is paid by the facilitator (Coinbase CDP / PayAI). - The request is retried with the signature → 200 + data +
PAYMENT-RESPONSE(tx hash).
Use a dedicated agent wallet with a few USDC. Never your main wallet.
Reading the output
direction:1supportive,-1negative,0unclear.confidence(0..1) = source tier × severity/novelty × graph-path weight — a ranking signal, not a probability.corroboration.count= independent sources. Primary sources (SEC, Fed, USGS) originate events; media only corroborates.pathshows the exposure route, e.g.nat.quake → facility:TSMC-Fab18 → company:TSM → company:NVDA.universe_versionis stamped on every response for reproducible backtests.
Also available as
- MCP server (Claude Code, Cursor, OpenClaw):
{"mcpServers": {"degenscan-intel": {"url": "https://intel.degenscan.io/mcp"}}} - Agent skill:
npx skills add tradewr333-lgtm/degenscan-intel - TypeScript/JavaScript:
npm i @degenscan/intel - Public metrics (paying wallets, tx hashes, operator wallets excluded):
https://intel.degenscan.io/v1/metrics
Limits
US-equity-heavy coverage (top-100 by volume + indices), 15 crypto, main commodities/FX/rates — check universe(). No price data, no forecasts. Liquidations are not part of derivs_for. Sources marked best-effort in sources() may go quiet.
MIT © Marbella Collins LLC · contact@degenscan.io
Metadata
Release files for degenscan-intel 0.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| degenscan_intel-0.2.0.tar.gz | 8.4 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| degenscan_intel-0.2.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 17.8 kB
Release files / degenscan_intel-0.2.0.tar.gz
| Download URL | degenscan_intel-0.2.0.tar.gz |
|---|---|
| Size | 8.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
9beaf0d5caa457bdd1767831c87307e2b89e13c5be084626f1b60137c6b805a2
|
|
BLAKE2b-256 checksum How to use checksums |
eee32284b37680d77515d69428a1142aa470ccab5a8e086ed0f9ceef571969e7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.2
|
Release files / degenscan_intel-0.2.0-py3-none-any.whl
| Download URL | degenscan_intel-0.2.0-py3-none-any.whl |
|---|---|
| Size | 9.4 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
a3b5d7c0a49ad9774f3a1b621528fd8c8f8b812da9facf0e6ee63a5846237fee
|
|
BLAKE2b-256 checksum How to use checksums |
57670e42c7f15830565227d8ee4ef865ec4e3e459c73a76343699f7c9d728b6e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.2
|