
Vibe-Trading MCP Server: Setup for Claude Code, Cursor and Other Apps
Gives Claude, Cursor and other assistants market data, backtesting and finance research tools.
Vibe-Trading is a finance research toolkit that Claude, Cursor and other assistants can use. It pulls market data and filings, runs backtests, tests trading signals, analyzes your trade journal and can read a broker account through read-only connectors. It runs on your computer, and core research works without a key. It runs strategy code for backtests and can see broker data, so use paper or read-only profiles.
Good for
- Retail traders who want an assistant to backtest an idea before risking money
- Quant researchers who test factors and signals with real data
- Investors who follow US, Hong Kong and China A-share markets
- Anyone who wants an assistant to review their own trade history
Works in
12 should work
Tap an app to see how to set it up and any known problems.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Client does not support this transport.
Client does not support this transport.
Install Vibe-Trading MCP
Claude Code
- 1
Copy this prompt.
Add the Vibe-Trading MCP server to Claude Code. Run this command: claude mcp add vibe-trading -- uvx --from vibe-trading-ai vibe-trading-mcp Then check that it loaded and tell me if I need to sign in or restart.
- 2
Paste it into Claude Code. It shows the change and asks before running it.
Requires uv (Python).
Read-only mode: No read-only mode for the whole server. Do not select a broker connection, and keep the research tools only.
The install box runs the package with uvx so you do not need to install it first; the README instead installs the package and then runs its server command. Needs Python 3.11 or newer. Add environment variables in your app's env block, not your shell: the app starts the server itself. To let backtests write into your own folder, set VIBE_TRADING_ALLOWED_RUN_ROOTS to that folder. For apps that connect by URL, the server can also run over HTTP on port 8900.
Built from the package data we hold. Not yet tested by us.
Tools
Gets price history for stocks and crypto
Fetches open, high, low, close and volume bars for stocks, crypto or both from the data source you pick.
Try: “Get daily prices for AAPL.US and BTC-USDT for the last year.”
Network ·
get_market_dataTurns a company name into a ticker
Finds candidate symbols for a name or fragment, in the project's format for A-shares, Hong Kong, US and other markets.
Try: “What is the ticker for Tencent?”
Network ·
search_symbolRanks a market by movers or activity
Screens a market and returns the top names by a metric, such as biggest movers or most traded.
Try: “Show today's top 10 US movers.”
Network ·
screen_marketShows a company profile and estimates
Returns valuation figures, analyst targets, estimates and ownership for a US, Hong Kong or UK listing.
Try: “Give me the profile of NVDA.US.”
Network ·
get_stock_profileGets income, balance and cash flow figures
Returns financial statements or key indicators per period, newest first, for A-share, US and Hong Kong stocks.
Try: “Show the last four quarters of income statements for MSFT.US.”
Network ·
get_financial_statementsGets fundamentals as daily panels
Returns fundamentals such as return on equity or net income as daily tables aligned by filing date, with point-in-time mode on by default.
Try: “Get ROE for these five stocks as a daily panel.”
Network ·
get_fundamentalsLists SEC filings or reported figures
Returns recent 10-K, 10-Q and 8-K filings for a US company, or its reported financial data.
Try: “List recent SEC filings for TSLA.”
Network ·
get_sec_filingsShows which funds hold a US stock
Reads Form 13F filings from SEC EDGAR to show a manager's portfolio or the holders of a ticker, with quarter-over-quarter changes.
Try: “Which funds increased their position in AAPL last quarter?”
Network ·
get_institutional_holdingsShows what is inside an ETF
Returns the holdings of an ETF or fund with weights and values, or searches funds by ticker, name or theme.
Try: “What are the top holdings of SPY?”
Network ·
etf_holdingsGets the options chain for a stock
Returns calls and puts for one expiration with strike, bid, ask, volume, open interest and implied volatility.
Try: “Show the AAPL options chain for the next expiry.”
Network ·
get_options_chainGets recent headlines for a stock
Returns recent financial news headlines for A-shares, US and Hong Kong stocks.
Try: “Latest news for 00700.HK.”
Network ·
get_stock_newsGets a US economic indicator from FRED
Returns dated values for series such as CPI, unemployment, GDP or the federal funds rate.
Try: “Plot the US unemployment rate since 2020.”
Network ·
get_macro_seriesReads Polymarket event odds
Reads public Polymarket data where a price is the market-implied probability of an event. It is read-only.
Try: “What are the odds that the Fed cuts rates in December?”
Network ·
prediction_marketFinds finance papers and summarizes them
Searches arXiv and OpenAlex for finance papers, or reads one into a short brief for building a factor.
Try: “Find recent papers on momentum factors.”
Network ·
research_papersScores text or shows crypto fear and greed
Scores text from bearish to bullish, or returns the crypto Fear and Greed Index.
Try: “Score the sentiment of this earnings headline.”
Network ·
sentimentShows the crypto order book
Returns a read-only snapshot of bids and asks from OKX or Binance with the spread and depth imbalance.
Try: “Show the BTC-USDT order book on Binance.”
Network ·
orderbook_depthCalculates RSI, MACD and other indicators
Computes RSI, MACD, Bollinger Bands, SMA and EMA plus volume statistics for a symbol.
Try: “Compute RSI and MACD for NVDA.US.”
Network ·
technical_indicatorsSpots chart patterns
Detects patterns such as head and shoulders, double tops, triangles and channels in price files from a run folder.
Try: “Look for chart patterns in the saved prices.”
Files ·
pattern_recognitionShows money flowing in or out of a stock
Returns net inflow by order size for A-share, Hong Kong and US stocks.
Try: “Is big money flowing into 600519.SH?”
Network ·
get_fund_flowGets the A-share top traders board
Returns the Chinese dragon-tiger board for a trade date, from Eastmoney.
Try: “Show yesterday's dragon-tiger board.”
Network ·
get_dragon_tigerShows Stock Connect inflows into A-shares
Returns the latest net inflow and recent history for Shanghai and Shenzhen Connect.
Try: “How much did northbound money buy this week?”
Network ·
get_northbound_flowShows Stock Connect inflows into Hong Kong
Returns the latest net buying and recent history into Hong Kong stocks.
Try: “Show southbound flows for the last month.”
Network ·
get_southbound_flowGets margin-trading balances for an A-share
Returns daily financing and securities-lending balances for one stock.
Try: “Show margin balances for 000001.SZ.”
Network ·
get_margin_tradingGets large A-share block trades
Returns recent block trades with price, amount, premium or discount and broker seats.
Try: “List recent block trades in 600519.SH.”
Network ·
get_block_tradesGets the shareholder count history
Returns the number of holders per reporting period for an A-share and the change from the last period.
Try: “How has the shareholder count of 000001.SZ changed?”
Network ·
get_shareholder_countGets share unlock dates
Returns the unlock schedule of one A-share, or a calendar of upcoming unlocks for the market.
Try: “Which stocks have big unlocks next month?”
Network ·
get_lockup_expiryLooks up A-share sectors and concepts
Lists the industry and concept boards a stock belongs to, or ranks boards by performance.
Try: “Which concept boards is 300750.SZ in?”
Network ·
get_sector_infoGets broker reports and forecasts
Returns recent sell-side reports for an A-share with ratings and per-year earnings forecasts.
Try: “What do analysts forecast for 600519.SH?”
Network ·
get_research_reportsGets NAV history for Chinese funds
Returns unit and accumulated net asset value and daily growth for off-exchange funds.
Try: “Show the NAV history of fund 110011.”
Network ·
get_fund_navGets China economic indicators
Returns series such as GDP, CPI, PMI, money supply, trade and industry data.
Try: “Show China PMI for the last two years.”
Network ·
get_cn_macro_seriesSearches Chinese company announcements
Searches filings for A-share, Hong Kong and fund announcements such as reports, dividends and buybacks.
Try: “Find buyback announcements from BYD.”
Network ·
get_cn_announcementsSearches broker research
Searches sell-side research by title, broker, industry and rating, with a highlighted excerpt.
Try: “Search broker reports on solid-state batteries.”
Network ·
search_broker_reportsScreens A-shares with a plain-language question
Runs a natural-language query on iWenCai, a Chinese stock screener. Chinese phrasing works best.
Try: “Find A-shares with rising profit and low debt.”
Network ·
iwencai_searchSearches a paid data marketplace
Finds data tools on the QVeris marketplace. Searching is free and shows expected cost.
Try: “Search QVeris for a tool that gives bond yields.”
Network ·
qveris_searchShows a marketplace tool's inputs
Fetches the full input details of QVeris tools before you run one. Inspection is free.
Try: “Inspect the QVeris tool you found.”
Network ·
qveris_inspectRuns a marketplace data tool
Runs one QVeris tool. The tool warns that it may be billable.
Try: “Run that QVeris tool for the 10-year yield.”
Network ·
qveris_executeSearches the web
Searches with configured providers and free engines and returns titles, links and snippets. No key is needed.
Try: “Search for news about the Fed meeting.”
Network ·
web_searchReads a web page as Markdown
Fetches a page and strips ads and navigation to leave clean text.
Try: “Read this earnings release page.”
Network ·
read_urlReads a PDF, including scans
Extracts text from a PDF and uses OCR for scanned pages.
Try: “Read this annual report PDF.”
Files ·
read_documentRuns a strategy backtest
Runs a vectorized backtest from a config file and a strategy script in a run folder, then saves results. The strategy script is code that runs on your computer.
Try: “Backtest a 20/50 day moving-average strategy on SPY for 5 years.”
Runs code ·
backtestSimulates one step of paper trading
Matches the orders you give against price bars in a simulated engine. It never places real orders.
Try: “Paper trade this order list for one day.”
Runs code ·
paper_trade_stepTests how well a factor predicts returns
Computes rank correlation, information ratio and quantile return spreads from prepared price and factor files.
Try: “Test whether this momentum factor predicts returns.”
Runs code ·
factor_analysisBrowses a library of ready-made signals
Lists the bundled alpha signals, shows one in detail or checks the library.
Try: “List the alphas in the Alpha Zoo.”
Read ·
alpha_zooBenchmarks signals on a stock universe
Runs one signal or the whole library over a universe such as the S&P 500 or CSI 300.
Try: “Benchmark alpha 101 on the S&P 500.”
Runs code ·
alpha_benchPrices an option and its Greeks
Calculates a Black-Scholes price with delta, gamma, theta and vega.
Try: “Price a call: spot 100, strike 105, 30 days, 25% volatility.”
Read ·
analyze_optionsAnalyzes a multi-leg option strategy
Returns cost, breakevens and profit at expiry for an option strategy, and what changes with price and volatility.
Try: “Analyze an iron condor on SPY.”
Read ·
analyze_options_payoffRuns a finance-math function
Runs a function from the tested finance library: option and bond math, risk measures like VaR, attribution, time-series tests and fund return maths.
Try: “Compute the 99% VaR for this return series.”
Runs code ·
quantlib_callMeasures returns with deposits and withdrawals
Computes time-weighted and money-weighted return for an account with cash flows during the period.
Try: “What was my true return given these deposits?”
Files ·
cashflow_performanceReads backtest results as structured data
Returns equity, trades, metrics or positions from a backtest run as JSON.
Try: “Show the trades from the last backtest.”
Files ·
read_run_artifactLists earlier backtest runs
Lists past runs, newest first, with the folder of each so results can be read without re-running.
Try: “List my recent backtests.”
Files ·
list_backtest_runsLists strategies and their evidence
Lists strategies from the Alpha Zoo and the strategy store with their evidence status.
Try: “What strategies are available?”
Read ·
list_strategiesFinds strategies that pass evidence filters
Ranks strategies by per-regime evidence from reproducible backtests.
Try: “Which strategies work in rising-rate regimes?”
Read ·
query_strategiesShows the evidence behind a strategy
Returns trade counts, coverage and Sharpe by market regime for one strategy.
Try: “Show the evidence for this strategy.”
Read ·
get_strategy_evidencePrepares a variation of a strategy
Prepares a child strategy folder from fresh evidence, keeping the signal logic, for you to edit and test.
Try: “Create a variant of this strategy for Hong Kong stocks.”
Write ·
adapt_strategyRebuilds the strategy evidence cache
Rebuilds a disposable cache of strategy evidence from backtest run files.
Try: “Refresh the strategy evidence.”
Write ·
refresh_strategy_evidenceWrites a file for a backtest
Writes content to a file, used to create the config and strategy script for a backtest. It is limited to allowed run folders.
Try: “Write config.json for a backtest.”
Write ·
write_fileReads a file
Reads a file by path.
Try: “Read signal_engine.py.”
Files ·
read_fileLists the built-in finance skills
Lists the finance skills with names and descriptions.
Try: “What finance skills do you have?”
Read ·
list_skillsOpens one finance skill
Loads the documentation of a named skill. Long skills come back as an outline you can drill into.
Try: “Load the options-strategy skill.”
Read ·
load_skillStarts a tracked research goal
Creates or replaces a research-only goal with a checklist, for long tasks that need audit.
Try: “Start a research goal to compare value and momentum.”
Write ·
start_research_goalShows the current research goal
Returns the current goal with its checklist and status.
Try: “Where are we on the research goal?”
Read ·
get_research_goalEdits the research goal
Changes the objective or criteria without losing evidence already added.
Try: “Narrow the goal to US large caps.”
Write ·
update_research_goalAdds evidence to the goal
Attaches a finding to the current research goal.
Try: “Add this backtest as evidence.”
Write ·
add_goal_evidenceMoves the goal to done, paused or blocked
Updates the goal status after an audit. Completing it needs a check of every criterion.
Try: “Mark the goal complete.”
Write ·
update_research_goal_statusLists multi-agent team presets
Lists teams of specialist agents such as an investment committee, quant desk or risk committee.
Try: “Which agent teams can I run?”
Read ·
list_swarm_presetsRuns a team of AI agents on a question
Runs a multi-agent team and streams progress. It needs an LLM key and uses your model provider.
Try: “Run the investment committee on NVDA.”
Runs code ·
run_swarmChecks a team run's progress
Returns status, task progress and token use for a swarm run.
Try: “How is the swarm run going?”
Read ·
get_swarm_statusGets a finished team run's report
Returns the final report and task summaries of a swarm run.
Try: “Show the result of the swarm run.”
Read ·
get_run_resultLists recent team runs
Lists swarm runs, newest first, and flags stale ones.
Try: “List my recent swarm runs.”
Read ·
list_runsRetries a failed team run
Starts a new run with the same preset and variables as a failed, stale or cancelled one.
Try: “Retry the failed swarm run.”
Runs code ·
retry_runCleans up stuck team runs
Marks runs whose process died as failed.
Try: “Clean up stuck swarm runs.”
Write ·
reap_stale_runsAnalyzes your broker trade history
Parses a trade export from 同花顺, 东方财富, 富途 or a generic CSV or Excel file and returns a trading profile and behavior diagnostics.
Try: “Analyze my trade journal in trades.csv.”
Files ·
analyze_trade_journalTurns your winning trades into rules
Derives a few if-then rules from your profitable trades in a journal.
Try: “Extract my shadow strategy from the journal.”
Read ·
extract_shadow_strategyBacktests those rules against your real trades
Runs a multi-market backtest of the rules and compares profit with what you actually made.
Try: “Backtest my shadow strategy.”
Runs code ·
run_shadow_backtestBuilds a report from the shadow backtest
Renders an HTML or PDF report with charts for a shadow account.
Try: “Render my shadow account report.”
Write ·
render_shadow_reportLists today's symbols matching your rules
Finds symbols that match the entry pattern of your shadow rules. The tool labels it research, not advice.
Try: “Which symbols match my shadow rules today?”
Network ·
scan_shadow_signalsLists broker connections you can pick
Lists connector profiles for brokers and paper trading accounts.
Try: “Which broker connections are available?”
Read ·
trading_connectionsPicks the broker connection to use
Sets the default connector profile for later trading calls.
Try: “Use the paper IBKR profile.”
Write ·
trading_select_connectionChecks the broker connection
Checks that a profile is set up and reachable. It never places orders.
Try: “Check my IBKR connection.”
Network ·
trading_checkReads your broker account
Reads account data from the selected profile.
Try: “Show my account balance.”
Network ·
trading_accountReads your broker positions
Reads positions from the selected profile.
Try: “What positions do I hold?”
Network ·
trading_positionsReads your open orders
Reads open orders from the selected profile. It does not place, cancel or change orders.
Try: “List my open orders.”
Network ·
trading_ordersGets a quote from your broker
Reads a price snapshot for a symbol from the selected profile.
Try: “Quote AAPL from my broker.”
Network ·
trading_quoteGets price history from your broker
Reads historical bars for a symbol from the selected profile.
Try: “Get one year of daily bars for AAPL from my broker.”
Network ·
trading_history
From the official docs
Permissions
high risk · 5 things to watch
High risk
The assistant gets many market-data and research tools plus a backtest tool that runs strategy code on your computer, and it can read your broker account once you pick a connection. Start with research-only use, keep broker connections on paper or read-only profiles, and treat results as research, not advice.
What it can do
- Read your data
- Create and change things
- Delete things
- Read-only mode
Watch out for
Backtests run strategy code on your computer
The backtest tool runs a strategy script the assistant writes. The README says generated code is sandboxed to allowed run folders; set VIBE_TRADING_ALLOWED_RUN_ROOTS to a folder you are happy to expose. github.com
It can read your brokerage account
After you select a connector, the trading tools read balances, positions and open orders. The tools do not place or cancel orders, but the assistant sees the data. github.com
One marketplace tool may cost money
qveris_execute is marked as possibly billable by the provider. Search and inspect are free. Do not approve execute calls you did not expect. github.com
Output is research, not financial advice
Backtests, signals and agent-team reports can be wrong or overfit. Check them before you trade. github.com
Shell tools are opt-in
The README says MCP shell tools need explicit opt-in at startup, and the MCP server loads the runtime .env. github.com
Security advisories
No published security advisories for the MCP server as of Nov 1, 2026.
3 other advisories are about the wider Vibe-Trading project, not this MCP server. See them on GitHub
Technical details
| Turning on read-only | No read-only switch for the whole server is documented. The trading_* tools only read, and the README says the connector layer is off and read-only by default. |
|---|---|
| Talks to | Public market-data and news sites (Yahoo Finance, Eastmoney, SEC EDGAR, FRED, Polymarket, OKX, Binance and others). Your broker connection, only after you select one. Your LLM provider when you run agent teams. QVeris, only if you run its marketplace tools (may be billable) |
| Sign-in and keys | Core research tools need no key for Hong Kong, US and crypto data. Other sources, agent teams and broker connections use API keys or logins you set in the runtime .env or the client env block. |
Questions
How many tools does it have?
The README says 84, a mix of market data, research, backtesting, strategy evidence, agent teams, trade-journal analysis and read-only broker tools. Some are mirrored from the project's own tool registry, so the exact number can change between versions.
Do I need an API key?
Not for core research with Hong Kong, US and crypto data. Agent teams need an LLM key, some data sources and broker connections need their own keys or logins, and the QVeris marketplace may charge.
Can it place real trades?
The MCP tools only read from brokers: account, positions, orders, quotes and history. The paper-trading tool never places real orders. The README says connectors are off and read-only by default.
Is this financial advice?
No. It is a research tool. Backtests and signals can be wrong, and the tool descriptions call some output research only.
Which setup should I use?
Install the Python package, then run its MCP command in your app. The install box uses uvx to do both; the README also shows a plain command after a pip install. It needs Python 3.11 or newer.
Does it work with Claude Desktop and Cursor?
Yes. The README gives a Claude Desktop config and says Cursor, Windsurf and other clients can run the same command. It also offers an HTTP mode on port 8900 for clients that connect by URL.
Information
updated Nov 2, 2026
- Publisher
- Community
- Package
vibe-trading-ai- Registry
- PyPI
- Transport
- stdio, Streamable HTTP
- Auth
- None
- License
- MIT
- Last commit
- Oct 30, 2026
- Open issues
- 11
Sources
Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.
Show all 3Hide
- Vibe-Trading README: MCP Plugin · github.com
- MCP server source · github.com
- vibe-trading-ai on PyPI · pypi.org