opensourceprojects.dev

A broadsheet for software that doesn't ask for your email

Claude Skills for structuring trading decisions, not outsourcing them
GitHub RepoViews78

Project Description

View on GitHub

A Trading Toolkit That Structures Your Decisions Instead of Making Them for You

You've probably seen a dozen AI trading tools that promise to tell you exactly what to buy and when. But what if the real problem isn't a lack of signals—it's a lack of process? Claude Trading Skills takes a different angle: it's a Claude Skills-based workflow toolkit that helps time-constrained individual investors structure market review, risk management, trade planning, and journaling, rather than outsourcing the actual buy/sell decision to an AI.

What It Does

Claude Trading Skills is built for investors whose core strategy is long-term investing, ETFs, and dividend stocks, with disciplined swing trading as a satellite approach when conditions are favorable. It's organized around operational workflows—each backed by a machine-readable manifest under workflows/ that names the exact skills, decision gates, and artifacts in order.

The toolkit ships with five recommended starting workflows: a 15-minute daily market check, a weekly long-term portfolio review, a swing-opportunity scan that only fires when risk is allowed, a trade-memory loop for recording and learning from closed trades, and a monthly performance review. Each workflow anchors to specific skills like market-breadth-analyzer, portfolio-manager, drawdown-circuit-breaker, position-sizer, and trader-memory-core. Some workflows need external data—Alpaca for portfolio review, FMP for screeners—while others run without any API at all. The project is MIT-licensed and explicitly not a signal service or a promise of profitability.

Why It's Cool

  • It inverts the usual AI trading pitch. The README is blunt about this: "The goal is not to outsource buy/sell decisions to AI." Instead, it structures the habits around decisions—review, risk, planning, journaling. That's a more honest use of AI for retail traders, and it's refreshing to see it stated so plainly.

  • The workflows are machine-readable manifests. This is a clever design choice. Each workflow YAML names the exact skills, decision gates, and artifacts in sequence, so the process isn't buried in prose—it's something a tool (or a Claude Skill) can actually execute and check against.

  • Risk gates come before opportunity. The swing workflow includes a drawdown-circuit-breaker and a pre-trade-discipline-gate alongside the screeners. You're managing risk before you're hunting for candidates, which is the opposite of how most trading tools are ordered.

  • Degraded fallbacks are documented. The portfolio review workflow requires Alpaca, but manual CSV is listed as a degraded fallback. That kind of honesty about what breaks without a data source is rare and useful.

  • It's a personal project first. The "first for self, open for others" stance means the workflows are ones the author actually uses, not features built to demo well. There's also a documentation site and a Japanese README, which suggests real care went into making it usable.

  • Clear boundaries. The README explicitly says it's not for fully automated trading, signal outsourcing, or short-term scalping. Knowing who a tool isn't for saves everyone time.

How to Try It

  1. Start at the repository: github.com/tradermonty/claude-trading-skills. The README is the entry point, and the documentation site has the fuller picture.

  2. Pick a single workflow based on your goal. If you just want a quick daily read on the market, start with market-regime-daily—it needs no API for the basic path. If you're reviewing a long-term portfolio, look at core-portfolio-weekly (Alpaca required, manual CSV as fallback).

  3. Open the corresponding manifest under workflows/ (for example, workflows/market-regime-daily.yaml) to see the exact skills, decision gates, and artifacts in order. These manifests are the real map of each workflow.

  4. If you want to see what output looks like before running anything, the repo includes sample runs—like examples/workflows/core-portfolio-weekly/sample-run/ and examples/workflows/swing-opportunity-daily/sample-run/.

  5. Read the disclaimer and privacy/security policies first. This is educational and process-improvement tooling, and the README is clear that all trading decisions, position sizing, and compliance remain yours.

Final Thoughts

Claude Trading Skills won't tell you what to buy, and it doesn't pretend to. What it offers is a structured, open-source process for the parts of trading most people skip—risk gates, journaling, and honest review—packaged as workflows you can actually follow in a short daily window. It's best suited to time-constrained individual investors who already have a long-term core and want discipline around a satellite swing strategy. If you're looking for signals, look elsewhere. If you're looking to build a better decision process, this is a solid place to start.


Follow @githubprojects for more developer tools and open source projects.

Back to Projects
Project ID: claude-trading-skillsLast updated: November 16, 2026 at 02:48 AM