Best Claude Skills for Stock Trading and Research (2026)

These are trading and market research skills that exist on GitHub today, checked one by one. For each you get what it does, how to install it, the star count, the license, and what it needs from your machine: scripts, API keys, network access, or broker credentials.

Nothing here is financial advice. These tools help you organise research and enforce your own rules. They do not know where prices are going. Star counts come from the GitHub API and were refreshed on 2026-11-12. If skills are new to you, start with what an agent skill is and Agent Skills explained. For skills outside trading, see the best Claude skills in 2026 hub list.

Quick comparison

Skill repo	Focus	Stars (2026-11-12)	License	Network / keys
Jakeschincariol/day-trader-skill	Day trading research	7,305	MIT	Public data, optional broker
tradermonty/claude-trading-skills	Equity research suite	3,791	MIT	FMP / FINVIZ / Alpaca keys for many skills
agiprolabs/claude-trading-skills	Quant, crypto, DeFi	527	MIT	Crypto data API keys
yennanliu/InvestSkill	US stock analysis prompts	283	MIT	None
alpacahq/alpaca-skills	Alpaca broker API	167	Apache 2.0	Alpaca keys
xvary-research/claude-code-stock-analysis-skill	Single-stock reports	21	MIT	SEC EDGAR, market data
JoelLewis/finance_skills	Wealth management, compliance, trading ops	225	MIT	Python reference scripts, no keys mentioned
roman-rr/trading-skills	Crypto perpetuals signals	127	Proprietary (per README)	Calls the vendor's signal API
Qyntc/CTK	Fork of tradermonty's suite	3	MIT	FMP keys for many skills
EshaRana17/claude-stock-research-skill	Research memo, risk sizing	New	MIT	None

The skills

tradermonty/claude-trading-skills

The most complete equity research set we found. Claude Trading Skills covers market breadth, uptrend participation, sector rotation, VCP and other screeners, technical analysis, earnings and economic calendars, macro regime detection, position sizing, a drawdown circuit breaker, backtesting and a trade journal. It also defines workflows that chain skills, for example a daily swing routine that only looks for setups when the risk gates allow it.

• Install: download a .skill package from the releases page and verify it against SHA256SUMS, or copy a skill folder into ~/.claude/skills/.
• Stars: 3,791 (stars as of 2026-11-12). License: MIT.
• Risk notes: hundreds of Python scripts. Many skills require a Financial Modeling Prep key, some use FINVIZ Elite or Alpaca, and some use yfinance. The README has a no-API-key starter path (breadth and uptrend analysis from public CSVs) that is a sensible first step.
• Docs: the project has its own documentation site with a getting-started guide, a skill catalog, per-skill guides, workflows, strategy playbooks and an FAQ, in English and Japanese.

Qyntc/CTK

CTK is a GitHub fork of tradermonty/claude-trading-skills: GitHub lists tradermonty's repo as its parent, it keeps the same "Claude Trading Skills" title and description, and its last push was in April 2026. Unless you have a specific reason to use the copy, prefer the upstream repo, which is actively maintained and publishes checksums for its releases.

• Stars: 3 (stars as of 2026-11-12). License: MIT.
• Risk notes: same shape as the upstream: many Python scripts and FMP API keys for several skills. Copies of popular repos are a common supply chain trick, so compare the code before trusting one. See AI agent supply chain attacks.

Jakeschincariol/day-trader-skill

The Day Trader skill bundles 11 skills: a chart scanner, news tracker, entry and exit timing, risk and budget rules, alerts, a congressional trading tracker, and a "time machine" that replays a setup thousands of times from recent bars to estimate odds. The README says nothing can place an order without your confirmation and paper trading is the default.

• Install: /plugin marketplace add Jakeschincariol/day-trader-skill, then /plugin install day-trader-skill@day-trader-skill
• Stars: 7,305 (stars as of 2026-11-12). License: MIT.
• Risk notes: 24 Python scripts that pull public data from Yahoo Finance, Nasdaq Trader, SEC EDGAR and the House Clerk, so expect network access. It includes a trader-broker skill; if you connect a broker, use paper trading and keep the confirmation step.

agiprolabs/claude-trading-skills

agiprolabs/claude-trading-skills has 68 skills leaning toward quantitative finance, crypto and DeFi: on-chain data, token analysis, strategy research and risk.

• Install: /plugin marketplace add agiprolabs/claude-trading-skills, then /plugin install trading-skills@agiprolabs-claude-trading-skills
• Stars: 527 (stars as of 2026-11-12). License: MIT.
• Risk notes: 119 Python scripts managed with uv, and API keys for services such as Birdeye, Helius and CoinGecko. Crypto tooling that touches wallets deserves extra care: never give a skill a private key.

yennanliu/InvestSkill

InvestSkill is a set of US stock analysis frameworks (fundamentals, valuation, earnings, technicals and more) written as plain prompts.

• Install: /plugin marketplace add yennanliu/InvestSkill, then /plugin install us-stock-analysis@invest-skill
• Stars: 283 (stars as of 2026-11-12). License: MIT.
• Risk notes: the README states there are no API keys and no runtime code; the skills are Markdown. That makes it low risk, but the analysis depends on data you paste in or your agent's own web search. Its alternative installer pipes a script from the internet into bash; use the plugin route or read install.sh first.

alpacahq/alpaca-skills

Alpaca's official skills teach agents to work with Alpaca's Trading and Broker APIs: market data, orders, paper trading, backtesting, account onboarding and funding transfers.

• Install: npx skills add alpacahq/alpaca-skills --skill alpaca-trading-backtest (or pick another skill)
• Stars: 167 (stars as of 2026-11-12). License: Apache 2.0.
• Risk notes: needs Alpaca API credentials and the Alpaca CLI. These skills can place real orders if you give them live keys. Start with paper keys and keep live keys out of the agent's environment.

xvary-research/claude-code-stock-analysis-skill

XVARY Stock Research adds /analyze, /score and /compare commands that produce a one-page thesis, a scorecard and risks for a ticker, backed by SEC EDGAR data.

• Install: /plugin marketplace add xvary-research/claude-code-stock-analysis-skill, then /plugin install xvary-stock-research
• Stars: 21 (stars as of 2026-11-12). License: MIT.
• Risk notes: Python tools that fetch SEC filings and market data over the network. It is the open part of a commercial research product, so expect links to the paid site.

JoelLewis/finance_skills

Finance Skills for Claude Code is aimed at financial services work rather than personal trading. Its README lists 91 skills in seven plugins: core return and time-value maths, wealth management (risk, valuation, portfolio construction), US securities compliance guidance, advisory practice, trading operations from order lifecycle to settlement, client operations and data integration. The repo's About line says 81, so the count depends on which you read.

• Install: /install JoelLewis/finance_skills per the README, or npx skills add JoelLewis/finance_skills
• Stars: 225 (stars as of 2026-11-12). License: MIT.
• Risk notes: the quantitative plugins include Python reference scripts; the compliance and operations plugins are guidance only. The README does not mention API keys or external services. It states the content is educational and not financial, legal, tax or investment advice.

roman-rr/trading-skills

roman-rr/trading-skills is a front end for a commercial crypto signal service. Its skills return a bullish or bearish call with a confidence score and reasoning for crypto perpetual futures, using Hyperliquid perpetuals data for coin selection, and a paid tier adds entries, stops and position sizes. The README reports its own hit rate; we have not checked it and do not repeat it here.

• Install: /plugin install roman-rr/trading-skills or npx skills add roman-rr/trading-skills
• Stars: 127 (stars as of 2026-11-12). License: the README describes it as proprietary with readable source; GitHub does not detect a standard licence.
• Risk notes: every signal is a call to the vendor's API (signals.x70.ai) with an API key, so your requests leave your machine. The README describes the bundled paper-trading bot as placing no real orders. Treat signals as one more opinion to check, not as instructions.

claude-stock-research-skill

Disclosure: built by the AgenticMart team.

claude-stock-research-skill (stock-research) structures research rather than generating signals. It produces:

• a research memo with an explicit bull case and bear case
• a technical snapshot from data you provide
• a position-size and risk calculator script that works from your account size, entry, stop and risk per trade
• a trade journal review that looks for patterns in your own past trades

What is distinctive: the calculator is standard-library Python with no network access, there are no API keys, it is MIT licensed, and it has been through our public security scan. It is new, so there is no usage history yet. It is educational and not financial advice.

• Install: git clone https://github.com/EshaRana17/claude-stock-research-skill ~/.claude/skills/stock-research
• Security report: scan results on AgenticMart
• License: MIT. Also listed on: the agent skills directory.

A note on crypto and DEX trading skills

Two skills above lean toward crypto. agiprolabs' set covers on-chain data, token analysis and DeFi using API keys for services such as Birdeye, Helius and CoinGecko, and roman-rr's signals draw on Hyperliquid perpetuals data. Decentralised exchanges settle trades from a wallet, so the stakes are different from a broker account: a leaked private key cannot be revoked like an API key. Keep research skills away from wallets entirely, never give a skill a private key or seed phrase, and sign any transaction yourself in a wallet you control.

Risk checklist for trading skills

Trading skills combine three risky things: scripts, credentials and money. Before installing one:

• Separate research from execution. A skill that reads data is very different from one that can place orders. Keep broker keys out of research skills.
• Use paper trading first. Alpaca and most brokers offer paper accounts. Use them until you trust the whole workflow.
• Scope your API keys. Use read-only data keys where possible, store them in environment variables, and never paste them into chat. Give broker keys read-only scope; see least privilege for agents.
• Check where data goes. Search the scripts for outbound URLs. Your watchlist and position sizes are private information.
• Pin versions. Install from a tagged release, and verify checksums when the maintainer publishes them.

Any AgenticMart listing can get a free automated scan, and each listing page shows its scan status. The scan checks a pinned commit for leaked secrets, vulnerable dependencies and network egress, and reads which domains and environment variables a listing declares; see how scanning works and the listings whose latest scan found no flagged issues. A scan is a malware and risk check, not a safety guarantee. For the general method, see how to write an Agent Skill and the SKILL.md format. Maintainers can submit a skill.

Using AI for trading responsibly (not financial advice)

This article is for information and education only. It is not financial, investment or tax advice, and AgenticMart is not a licensed adviser. Talk to a qualified professional before making investment decisions.

A few rules make these tools safer to use:

• Treat output as a draft, not a decision. Language models can misread numbers, use stale data and state wrong figures with confidence. Check every number against the primary source, such as the company's filings.
• Ask for the bear case. A model will happily build a bullish story for any ticker. Make it argue the other side and name what would prove the idea wrong.
• Size positions before you look for entries. Decide how much you can lose on a trade first. A position-size calculator is one of the most useful things a skill can provide.
• Keep a human in the loop. Do not let an agent place trades unattended. Every order should need your explicit confirmation.
• Backtests are not forecasts. Simulated results ignore slippage, fees and the fact that you tuned the rules to past data.
• Journal and review. The most reliable edge most people have is learning from their own mistakes. Use AI to review your journal, not to replace your judgement.

FAQ

Can a Claude skill trade stocks for me?

Some can connect to a broker. Alpaca's skills and the Day Trader skill's broker skill are examples. We recommend keeping them on paper trading and requiring confirmation for every order. Research-only skills cannot place orders at all.

Which trading skill needs no API keys?

InvestSkill is plain Markdown with no keys. The Day Trader skill uses free public data without signup. tradermonty's suite has a no-key starter path. The AgenticMart stock-research skill needs no keys or network access.

Is AI stock analysis reliable?

It is useful for organising research and spotting what you missed, not for predicting prices. Verify figures against filings and treat any conclusion as a starting point.

Is Qyntc/CTK the same as tradermonty's skills?

It is a GitHub fork of tradermonty/claude-trading-skills with 3 stars as of 2026-11-12 and no pushes since April 2026. Use the upstream repo, which is maintained and publishes release checksums.

Can I use these skills for crypto?

Some of them. agiprolabs' set covers on-chain data, tokens and DeFi, and roman-rr's skills return crypto perpetuals signals from a paid service. Keep wallets and private keys out of the agent's reach and sign transactions yourself.

Do I need Claude Code to use trading skills?

Most installs above use Claude Code plugins or npx skills add. Plain SKILL.md skills can also be uploaded to claude.ai as a ZIP under Customize > Skills with code execution on (Claude Help Center), but skills that run Python scripts with API keys are easier to control on your own machine.

Is this financial advice?

No. This is an educational overview of software tools. Consult a licensed professional for advice on your own situation.