trading-agents-skill
Moved to github.com/pyuan07/trading-skills
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/pyuan07--trading-agents-skill)HTML
<a href="/listings/pyuan07--trading-agents-skill"><img src="/api/listings/pyuan07--trading-agents-skill/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
1. Copy the skill folder into your skills directory:
| Tool | Location | |---|---| | Claude Code (all projects) | `~/.claude/skills/trading-agents/` | | Claude Code (one project) | `<project>/.claude/skills/trading-agents/` | | Codex | `<project>/.agents/skills/trading-agents/` |
git clone https://github.com/pyuan07/trading-agents-skill cp -r trading-agents-skill/skills/trading-agents ~/.claude/skills/
2. Give the skill a way to run its Python scripts. **Easiest: install [uv](https://docs.astral.sh/uv/getting-started/installation/)** and you're done. Each script declares its own dependencies, and `uv run` fetches them automatically the first time.
Or, without uv, create the skill's own venv once (Python 3.10+):
# macOS / Linux cd ~/.claude/skills/trading-agents python3 -m venv .venv && .venv/bin/python -m pip install -r scripts/requirements.txt
# Windows cd $HOME\.claude\skills\trading-agents python -m venv .venv; .venv\Scripts\python.exe -m pip install -r scripts\requirements.txt
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 127 lines
- > **📦 Moved:** this skill now lives at **[pyuan07/trading-skills](https://github.com/pyuan07/trading-skills)**. This repository is archived and will not receive updates.
- # trading-agents — a multi-agent stock analysis skill
- An agent skill for **Claude Code** (and other tools that load `SKILL.md` skills, such as Codex) that runs the [TradingAgents](https://github.com/TauricResearch/TradingAgents) methodology inside your coding agent:
- ```
- analysts (fundamentals · technical · news/sentiment)
- → bull vs bear debate
- → trader plan
- → risk debate (aggressive · conservative · neutral)
- → portfolio manager: BUY / ADD / HOLD / TRIM / SELL + size, stop, target
- ```
- It is **portfolio-aware**: give it your holdings and it judges the ticker against your real book (concentration, sector weights, correlation, margin, options exposure, your risk profile). It also keeps a **decisions ledger** that grades every past call against SPY at 30 and 180 days, and shows that track record to the agents the next time you analyze the same ticker.
- > ⚠️ **Decision support, not financial advice.** LLM output can be wrong, and market data can be stale or missing. Verify everything before trading. This skill never places orders.
- ## What you get
- - `reports/YYYY-MM-DD-TICKER.md`: the full report, with every stage's reasoning and a decision summary.
- - `reports/ledger.md`: one row per call, graded automatically as each horizon matures.
- - Deterministic Python for the number-crunching (indicators, FX, concentration, ledger math), so the LLM spends its tokens on judgment.
- **See a full example:** [`docs/sample-report-MSFT.md`](docs/sample-report-MSFT.md). It's a complete run in which the analysts debate, the risk team argues over position size, and the portfolio manager trims an oversized position back under the 20% cap.
- ## Install
- 1. Copy the skill folder into your skills directory:
- | Tool | Location |
- |---|---|
- | Claude Code (all projects) | `~/.claude/skills/trading-agents/` |
- | Claude Code (one project) | `<project>/.claude/skills/trading-agents/` |
- | Codex | `<project>/.agents/skills/trading-agents/` |
- ```bash
- git clone https://github.com/pyuan07/trading-agents-skill
- cp -r trading-agents-skill/skills/trading-agents ~/.claude/skills/
- ```
- 2. Give the skill a way to run its Python scripts. **Easiest: install [uv](https://docs.astral.sh/uv/getting-started/installation/)** and you're done. Each script declares its own dependencies, and `uv run` fetches them automatically the first time.
- Or, without uv, create the skill's own venv once (Python 3.10+):
- ```bash
- # macOS / Linux
- cd ~/.claude/skills/trading-agents
- python3 -m venv .venv && .venv/bin/python -m pip install -r scripts/requirements.txt
- ```
- ```powershell
- # Windows
- cd $HOME\.claude\skills\trading-agents
- python -m venv .venv; .venv\Scripts\python.exe -m pip install -r scripts\requirements.txt
- ```
- ## Use
- Work from any folder you like. That folder becomes the skill's workspace: the skill reads your holdings from it and writes `work/` and `reports/` into it.
- ```
- analyze NVDA
- should I trim MSFT?
- deep dive on 7203.T ← "deep dive" / "full pipeline" runs every stage as its own subagent
- review my portfolio ← screens all holdings, then you pick which ones get a full run
- ```
- ### Optional: your portfolio
- Put these files in your working folder. Templates are in `skills/trading-agents/examples/`.
- - **`holdings.csv`**: one row per position. Only `ticker` and `qty` are required:
- ```csv
- ticker,name,qty,price,cost,market_value,currency,account,margin
- AAPL,Apple Inc.,50,230.00,180.00,11500.00,USD,brokerage,no
- 7203.T,Toyota Motor,100,2800,2500,280000,JPY,brokerage,no
- ```
- - Use Yahoo Finance symbols (`7203.T`, `ES3.SI`, `0700.HK`). OCC option symbols (`SPY270115P00500000`) are detected automatically and treated as exposure context.
- - Several files (`holdings-ira.csv`, `holdings-taxable.csv`) are merged.
- - **moomoo / Futu users:** drop the position export (`持仓-<account>(<id>)-<timestamp>.csv`) into the folder as-is. It's read natively.
- - **`portfolio-profile.json`**: `risk_tolerance`, `max_position_pct`, `margin_comfort`, `horizon`, and free-text `notes` that the portfolio manager must respect.
- Without either file the skill still works. It just analyzes the ticker without reference to a portfolio.
- ### moomoo users: live positions and extra research (optional)
- If the **moomoo MCP** is connected to your agent, the skill uses it automatically, and you don't need a CSV:
- | From moomoo | Used for |
- |---|---|
- | `account_positions` (all your securities accounts) | Live holdings, including short positions and options, instead of a CSV export |
- | `quote_research_analyst_consensus` | Analyst rating mix and target-price range, passed to the fundamentals analyst |
- | `quote_short_interest` | Short-interest trend and days to cover, passed to the fundamentals and news analysts |
- | `quote_unusual_capital` (optional) | Unusual capital and short-selling activity, passed to the news analyst |
- The skill only **reads** from moomoo. Its instructions forbid calling any order, trade, cancel or modify tool. If the MCP isn't connected, or a market isn't covered by your data permissions, the skill quietly falls back to `holdings.csv` and Yahoo.
- Price history and indicators still come from Yahoo, so the skill works the same for everyone, moomoo customer or not.
- ### Run depth and cost
- | Depth | What runs | When |
- |---|---|---|
- | **lite** (default) | Scripts, plus subagents for 2 analysts and the bull/bear debate. Trader, risk and PM are written inline by the orchestrator. | Everyday use |
- | **full** | Every stage is a separate subagent | Ask for a "full pipeline" or "deep dive" |
- ## Data sources
- - **Yahoo Finance via [yfinance](https://github.com/ranaroussi/yfinance)**: prices, fundamentals, headlines, FX. It's free and needs no key. It is an unofficial API, so it's subject to Yahoo's terms and occasional breakage, and the skill falls back to web lookups when it fails.
- - **Web search** for context such as why a stock moved and breaking news.
- - **moomoo MCP** (optional): live positions and research data, as described above.
- - **Alpha Vantage MCP** (optional): used sparingly if you have it connected.
- All values are normalized to **USD**, and the benchmark is **SPY**.
- ## Development
- ```bash
- cd skills/trading-agents/scripts
- pip install -r requirements-dev.txt
- pytest -q # offline: yfinance is faked in tests
- ```
- ## Credits and license
- This project adapts the methodology and agent roles of [TradingAgents](https://github.com/TauricResearch/TradingAgents) (Tauric Research, Apache-2.0). See [NOTICE](NOTICE). It is released under the [Apache License 2.0](LICENSE).
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.