n0brains-cli 1.2.0

Token-efficient CLI for the n0brains crypto intelligence API — built for trading agents.

n0brains CLI

Token-efficient command-line access to the n0brains crypto intelligence API — built for trading agents, scripts, and humans. Every command prints raw JSON to stdout, so it composes with jq, pipes into an LLM, or feeds a bot directly. Zero third-party dependencies (Python 3.8+ stdlib only); the optional stream command needs websockets.

Why a CLI (vs the MCP server)

n0brains also ships an MCP server for Claude / Cursor / Cline. The CLI is the better fit when you're driving a bot or shell pipeline: plain JSON with no protocol envelope is more token-efficient, it composes with Unix tools, and it runs anywhere curl does. Same data, both surfaces.

Install

pip install n0brains-cli          # once published
# or run the single file directly:
python3 n0brains_cli.py --help

Auth

export N0BRAINS_API_KEY=intel_sk_...   # get one at https://n0brains.com
# or pass --api-key on any command. Public endpoints (proof) work without a key.

Commands

Command	What it returns
signals [--proven-only] [--asset BTC] [--type hack] [--limit N]	Active classified signals
signal <id>	One signal with full context
similar <query>	Semantic similarity search (pro)
levels <coin>	Support/resistance levels
liquidation-map <coin>	Estimated liquidation-cluster zones
correlation [coin] [--vs BTC] [--coins BTC,ETH,SOL]	Correlation matrix or one coin's corr+beta
rotation	Altseason/rotation read (into_alts / into_btc / neutral)
macro	Daily BTC/ETH macro bias
narrative <coin>	Narrative context
trust <coin>	Trust/rug score
manipulation <coin>	Manipulation score
proof [--equity]	Live forward-return proof per signal type (+ equity curve)
performance [--days 30] [--tier N]	Backtest stats
plan <coin>	Assembled trade plan — direction, entry, strongest target, stop, R:R, options context + warnings (pro)
rank [--coins BTC,ETH,SOL]	Cross-asset ranking — which coin is the best trade now (pro)
state [coin]	One-call per-coin snapshot (price + market-state + levels + regime); omit coin for the unified market state (pro)
regime	Market regime read — risk-on/off score (pro)
liquidity	Liquidity map — net liq, stablecoin dry powder, OI, taker flow (pro)
flows	Capital-flow map — rotation, cross-asset OI, institutional posture (pro)
market-map	One-call cross-asset market map (regime + liquidity + flows) (pro)
discovery	Emergent-edge discovery — corroboration patterns with live forward-return edge (pro)
stream	Live signal stream over WebSocket (pro)

Examples

# Only proven-edge signals, as JSON
n0brains signals --proven-only --limit 10

# Is it altseason? one-line answer
n0brains rotation | jq -r .regime

# Where are BTC liquidation magnets?
n0brains liquidation-map BTC | jq '.densest_long_liq_zone, .densest_short_liq_zone'

# Feed proven-edge signals into a bot loop
while n0brains signals --tradeable-only --limit 5 | jq -c '.signals[]'; do sleep 60; done

# Live stream (needs: pip install websockets)
N0BRAINS_API_KEY=... n0brains stream | jq -c 'select(.conviction=="strong")'

Exit code is 2 on an API error (so scripts can detect failures), 0 on success.

For agent frameworks

See SKILL.md for a capability description you can drop into an agent framework (the same tools are also exposed via MCP and REST).