skill
claude-skills-for-crypto-quant
Production-grade Claude Code skills for the crypto/quant trading vertical.
Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/MrMaca11an/claude-skills-for-crypto-quant itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Oct 23, 2026Claude Skills for Crypto & Quant
Built and open-sourced by the team behind Millennium AI
A curated pack of Claude Code skills for building production AI systems in crypto/quant trading.
Everyone's "awesome skills" list is full of generic web-dev and debugging skills. Nobody ships a serious pack for the crypto/quant vertical. This is that pack — the engineering discipline we actually use to build an AI crypto-trading platform, packaged so your agent inherits it too.
These skills don't contain any trading strategy, signal, or secret. They encode how to build well in this domain — bias-free backtesting, risk discipline, market microstructure literacy, and the data/LLM infrastructure that powers it.
What's inside
🧪 Trading discipline
| Skill | What your agent learns |
|---|---|
backtesting-discipline | No lookahead, realistic fees/slippage, regime splits, the metrics that actually matter |
strategy-optimization | Walk-forward, out-of-sample gates, plateau-over-peak, anti-overfitting |
risk-management | Position sizing, drawdown gates, correlation, leverage realism |
market-microstructure | Order flow / CVD, open interest, funding, liquidations |
crypto-liquidity-flow | ETF flows, derivatives, on-chain, dominance/rotation as one picture |
🏗️ Infrastructure
| Skill | What your agent learns |
|---|---|
llm-integration | Streaming, RAG pipelines, model selection, cost optimization |
senior-prompt-engineer | Prompt design, RAG optimization, multi-model orchestration |
redis-patterns | Caching, pub/sub, streams, rate limiting, distributed locks |
qdrant-vector-db | Semantic search, collections, payload filtering, RAG retrieval |
advanced-sql-timescale | Window functions, continuous aggregates, hyperfunctions for time-series |
fastapi-best-practices | Async endpoints, SSE streaming, error handling, auth |
websocket-realtime | SSE/WebSocket streaming, reconnection, heartbeat, scaling |
Install
Skills are just folders with a SKILL.md. Drop them where Claude Code looks
for skills:
# project-level (recommended): available in this repo only
git clone https://github.com/MrMaca11an/claude-skills-for-crypto-quant.git
cp -r claude-skills-for-crypto-quant/skills/* .claude/skills/
# or user-level: available everywhere
cp -r claude-skills-for-crypto-quant/skills/* ~/.claude/skills/
Claude Code auto-discovers them. Each skill's description tells the agent
when to use it — so the relevant discipline activates automatically when you
ask it to write a backtest, design a strategy, or build a data pipeline.
Want just one? Copy a single folder:
cp -r claude-skills-for-crypto-quant/skills/backtesting-discipline ~/.claude/skills/
How skills work (30-second version)
A skill is a markdown file with frontmatter:
---
name: backtesting-discipline
description: Enforces rigorous, bias-free backtesting... Use when writing or
reviewing any backtest, strategy simulation, or performance report.
---
# Backtesting Discipline
...the actual guidance the agent follows...
Claude reads the description to decide when a skill is relevant, then pulls in
the body as guidance. No code to run, no setup — it's knowledge your agent gains.
Pairs well with
- backtest-truth — a
linter that enforces
backtesting-disciplinein CI. - autonomous-agent-md — the operating manual for running your agent autonomously.
Contributing
PRs welcome — especially new crypto/quant skills (options, perps funding arbitrage, on-chain analytics). Keep each skill short and actionable; the value is in the discipline, not the word count.
⚠️ Not financial advice. These skills teach engineering discipline, not profitable strategies.
License
MIT. Curated and open-sourced by the team behind Millennium AI.