QuantPower · Quanty MCP · Setup and tools

Quanty MCP setup and tools

Connect QuantPower to Claude, ChatGPT, Codex, Gemini, Grok, Cursor or Perplexity, and see all 28 tools your AI gets. This list comes from the server's own tool list.

Server address

Use this one address in every AI app.

https://api.quantpower.tech/mcp

The first time, your AI app opens the QuantPower sign-in page. Sign in and click Allow. Remove access at any time from Connect AI in the QuantPower app.

Transport: MCP streamable HTTP. Sign-in: OAuth 2.1 with PKCE; the AI app registers itself. Apps that take a header (Claude Code, Codex, Gemini CLI, Cursor) can instead use a token you create in Connect AI in the QuantPower app: Authorization: Bearer qpa_…

Connect your AI

Open your app for its steps.

Claude claude.ai, desktop, phone, Claude Code

Works on: claude.ai and the Claude desktop app. Once added, it is in the Claude phone app too. Claude Code on the command line.
Plan: Free (one custom connector), Pro, Max, Ultra, Team and Enterprise.
Sign-in: QuantPower sign-in page (OAuth). Claude Code can also take a token.

claude.ai or the desktop app

Open Customize → Connectors (claude.ai/customize/connectors).
Click + Add, then Add custom connector.
Name it QuantPower, paste the server address above, click Continue.
Under Authentication choose Sign in now. Under OAuth client choose Register automatically. Click Add.
Sign in to QuantPower on our page and click Allow.
In a chat, tap +, then Connectors, and turn QuantPower on.

Claude phone app

Add it once on claude.ai or the desktop app. It then shows in the Claude app on iPhone and Android, under the same account.

Team or Enterprise

An Owner adds it in Organization settings → Connectors (Add, Custom, Web) with the same address. Each member then clicks Connect in Customize → Connectors and signs in with their own QuantPower account.

Claude Code

Terminal

claude mcp add --transport http quantpower https://api.quantpower.tech/mcp
claude mcp login quantpower

Or type /mcp inside Claude Code and choose quantpower to sign in.

Source: Claude Help Center: custom connectors

Source: Claude Code docs: MCP

ChatGPT chatgpt.com on the web

Works on: chatgpt.com in a browser. OpenAI's help page says custom MCP apps are web only, not in the phone app.
Plan: OpenAI lists Business, Enterprise and Edu for full MCP apps, and Pro for read tools in developer mode. Your account and workspace settings decide.
Sign-in: QuantPower sign-in page (OAuth).

Open chatgpt.com/plugins, select +, then Add custom MCP server.
Enter a name (QuantPower) and a short description.
Under Connection, paste the server address above.
Choose OAuth sign-in, read the risk note and confirm.
Select Create as a plugin, review the tools, sign in to QuantPower and click Allow.

On a Business workspace, an admin turns on developer mode first.

Source: OpenAI: connect a custom MCP server in ChatGPT

Source: OpenAI Help: plans for MCP apps

Codex command line and IDE extension

Works on: Codex CLI and the Codex IDE extension. Both read the same settings file.
Plan: Your Codex sign-in. OpenAI's MCP page names no plan.
Sign-in: QuantPower sign-in page (OAuth), or a token in an environment variable.

Add this to ~/.codex/config.toml:

Text

[mcp_servers.quantpower]
url = "https://api.quantpower.tech/mcp"

Then sign in:

Terminal

codex mcp login quantpower

To use a token instead, add bearer_token_env_var = "QUANTPOWER_TOKEN" under the same heading and put your token in that environment variable.

Source: OpenAI Codex docs: MCP

Gemini Gemini CLI, and the Gemini app in the US

Works on: Gemini CLI on your computer. The Gemini app adds custom apps only on gemini.google.com, for some accounts (below).
Plan: Gemini CLI: your Gemini CLI sign-in. Gemini app: personal Google account, 18 or over, in the US, in English, with Keep Activity on. Not with a work or school account.
Sign-in: QuantPower sign-in page (OAuth). Gemini CLI can also take a token.

Gemini CLI

Terminal

gemini mcp add --transport http quantpower https://api.quantpower.tech/mcp

Then type /mcp auth quantpower inside Gemini CLI. Sign-in needs a browser on the same computer.

Gemini app (US accounts only)

On a computer, open gemini.google.com, then Settings → Connected Apps (or Personal Intelligence → Connected Apps).
Under Custom apps, click Add a custom app and paste the server address above.
Click Next, sign in to QuantPower and click Allow.
In a prompt, type @ and pick QuantPower.

Once added on the web, it also works in the Gemini phone app. It cannot be added from the phone. Outside the US, use Gemini CLI.

Source: Gemini CLI docs: MCP servers

Source: Gemini Apps Help: custom apps

Grok grok.com

Works on: grok.com in a browser. xAI's pages do not cover adding it from the Grok phone app or X.
Plan: xAI says connectors are open to all Grok users. On Grok Business and Enterprise, a team admin adds the connector first in console.x.ai.
Sign-in: QuantPower sign-in page (OAuth).

Open grok.com/connectors.
Click New Connector, then Custom.
Paste the server address above, sign in to QuantPower and click Allow.

Source: xAI docs: Grok connectors

Cursor desktop editor

Works on: The Cursor editor on your computer.
Plan: Cursor's MCP page names no plan. Team and Enterprise admins can limit MCP servers.
Sign-in: QuantPower sign-in page (OAuth with PKCE), or a token in a header.

Add this to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project). Cursor opens the QuantPower sign-in page when it first connects.

JSON

{
  "mcpServers": {
    "quantpower": {
      "url": "https://api.quantpower.tech/mcp"
    }
  }
}

Manage it later from the Customize page in Cursor's sidebar.

Source: Cursor docs: MCP

Perplexity perplexity.ai

Works on: perplexity.ai. Perplexity's help page does not say which devices can add one.
Plan: Perplexity's help page does not list plans. In an organization, an admin turns on Allow members to add custom connectors.
Sign-in: QuantPower sign-in page (OAuth). Perplexity has no field for a token header.

Open Plugins in the sidebar (or Customize → Connectors).
Select Add custom, then Add MCP connector, then Remote.
Name it QuantPower and paste the server address above.
Under Advanced, set Authentication to OAuth and Transport to Streamable HTTP.
Tick the acknowledgement and select Add.
Select the QuantPower card, sign in to QuantPower and click Allow.

Source: Perplexity Help Center: custom remote connectors

Any other MCP app

Give the app the server address and choose streamable HTTP. The app finds the sign-in by itself.

All 28 tools

Today's orders with status, quantity, filled quantity and prices. Read-only; orders cannot be placed, changed or cancelled here.

Arguments

status text, optional, default all · Which orders to list. Values: all, open, executed, rejected, cancelled, amo

Arguments

algo_id text, required · The algo's id from get_algos.
date text, optional · YYYY-MM-DD (optional; today if left out).

Market data

Reads quotes, candles, option chains, scanners, market hours and QuantPower help.

`get_market_indices`

Reads data

Market indices

Connected index values. Preserve the source and timestamp; receipt time is not quote time.

No arguments

`search_scrip`

Reads data

Search scrips

Find stocks, futures and options by symbol or company name. Returns the instrument id the other tools need.

Arguments

query text, required · Symbol or name, e.g. RELIANCE or Reliance Industries.
segment text, optional · Market segment. Values: NSECM, NSEFO, NSECD, NSECO, MCXFO, BSECM, BSEFO, BSECD, BSECO

`get_quote`

Reads data

Quote

Last price, change, open, high, low, close, volume and circuit limits for one instrument. Read-only.

Arguments

instrument text, required · Instrument id from search_scrip, e.g. NSECM:2885 (segment:exchange instrument id).

`get_candles`

Reads data

Candles

Price candles (open, high, low, close, volume) for one instrument and timeframe. Times are Unix seconds.

Arguments

instrument text, required · Instrument id from search_scrip, e.g. NSECM:2885 (segment:exchange instrument id).
timeframe text, required · Candle size. Values: 1m, 3m, 5m, 15m, 30m, 1h, 1d, 1w
from text, optional · Start, ISO date-time (optional), e.g. 2026-10-23T09:15:00+05:30.
to text, optional · End, ISO date-time (optional).

`get_option_chain`

Reads data

Option chain

Option chain for an index or stock with strikes, last prices, open interest, IV and delta. Read-only.

Arguments

underlying text, required · Underlying symbol in capitals, e.g. NIFTY, BANKNIFTY, RELIANCE.
expiry text, optional · Expiry date YYYY-MM-DD (optional; nearest expiry if left out).

`run_scanner`

Reads data

Run a scanner

Run a market scanner now (top gainers, volume shock, RSI below 30 and others) and list the matching scrips.

Arguments

scanner_id text, required · Scanner id, e.g. top_gainers.
segment text, optional · Market segment. Values: NSECM, NSEFO, NSECD, NSECO, MCXFO, BSECM, BSEFO, BSECD, BSECO

`get_market_status`

Reads data

Market hours and holidays

Which markets are open now (equity, F&O, MCX, currency), today's hours, upcoming holidays and the next index expiries.

No arguments

`search_help`

Reads data

Search QuantPower help

Answer 'how do I…' questions from QuantPower's own help pages, with the page link.

Arguments

query text, required · The question in a few words, e.g. scalper one tap.

Calculate

Works on rules and legs you give it, and backtests them. Nothing is saved or placed.

`review_option_strategy`

Calculates

Review an option drawing

Validate explicit user-supplied option legs and sizing for review. Pure validation only: no saved-account access, save, enable or order. Ask for missing expiry and sizes; never invent rules.

Arguments

strategy object, required
- name text, required
- underlying text, required Values: NIFTY, BANKNIFTY, FINNIFTY, MIDCPNIFTY, SENSEX, BANKEX
- expiry text, required
- mul whole number, required
- legs list, required

`backtest_rules`

Calculates

Backtest your rules

Backtest the client's own rules on historical candles with the chart's rule engine (the same indicators, candle patterns and chart formations as the chart).

Full description

Backtest the client's own rules on historical candles with the chart's rule engine (the same indicators, candle patterns and chart formations as the chart). Use it when the client asks to backtest or test an indicator, candle-pattern or formation rule. strategy = {version: 1, name, side: "long"|"short", quantity (units), entry: {logic: "all"|"any", conditions: [...]}, optional exit (same shape), optional stopLoss / target ({kind: "percent", value} | {kind: "points", value} | {kind: "atr", period, multiplier}), optional trailingStop ({kind: "indicator", indicator, output} | {kind: "percent", value}), optional squareOff "HH:MM" (intraday), optional costs {slippagePct, perOrder}}. Conditions: {kind: "compare", left, op: above|below|cross-up|cross-down, right}, {kind: "direction", indicator, op: is-up|is-down|turns-up|turns-down|turns}, {kind: "change", operand, op: rises|falls}, {kind: "candle-pattern", id}, {kind: "chart-pattern", id}; an operand is {"kind": "price", "field": "close"} (open/high/low/close/volume), {"kind": "value", "value": 30} or {"kind": "indicator", "indicator": {"kind": "ema", "period": 20, "multiplier": 2}, "output": 0}; the client's own custom indicator is {"kind": "indicator", "indicator": {"kind": "custom", "custom": {"name", "lines": [{"formula", "label"}], "pane"}}, "output": line index} (formulas as in save_custom_indicator). Indicator kinds include ema, sma, rsi, macd, supertrend, supertrend-ema, supertrend-sma, vwap, bollinger, atr-trailing, chandelier, cpr-daily, auto-fib, smc-structure and 90 more; period 2-300, multiplier 0.1-10. Fills are at the next bar's open. Results are past results, not advice: give the key figures and say they are historical; never call a strategy profitable or tell the client to use it.

Arguments

instrument text, required · Instrument id as from search_scrip, e.g. NSECM:2885.
timeframe text, required · Candle size. Values: 1m, 3m, 5m, 15m, 30m, 1h, 1d, 1w
bars whole number, optional · How many recent closed bars (default 1000).
strategy object, required · The rules, as described above.

`run_backtest`

Calculates

Backtest a signal algo

Full description

Backtest the client's own entry / exit rules on one instrument with the app's backtest engine (the Backtest button: entry at the close of the bar the rule turns true, stop-loss / target inside a later bar, slippage, estimated charges) and return the report: trades, win %, net P&L, avg win / loss, reward-to-risk, expectancy, max drawdown (INR and %), drawdown duration, streaks, day-wise and monthly P&L. strategy = {instrument (search_scrip), tf 5m|15m|30m|1h|1D, side long|short, qty, entry: {left, op >|<|crosses_above|crosses_below, right}, exit (same shape, optional), sl_pct, target_pct, bars 100-2000, slippage, slippage_unit pts|pct}; an operand is a number or {"indicator": "price"} | {"indicator": "sma"|"ema"|"rsi", "period": 2-200} | {"indicator": "macd_hist"}. One of exit / sl_pct / target_pct is needed. Results are past results on historical data, not a promise or advice: give the key figures and say they are historical; never call a strategy profitable or recommend it. This is software, not a research analyst or adviser: it builds and checks the client's own rules and never recommends a strategy, a trade or says one will be profitable.

Arguments

strategy object, required · The rules, as described above.

Build

Checks an algo built from your own rules, or saves it switched OFF. Only you switch an algo on, in the app.

`list_templates`

Reads data

Algo templates

Ready option-algo templates (short straddle, short strangle, iron fly, iron condor, bull call / bear put spread, long straddle) in the saved-algo format, every rule written out, to start the client's own algo from. Read-only. This is software, not a research analyst or adviser: it builds and checks the client's own rules and never recommends a strategy, a trade or says one will be profitable.

No arguments

`build_strategy`

Checks an algo

Check an algo (not saved)

Check one option algo built from the client's own words with the app's own rules and return it normalised (legs in words, schedule, exits) or the exact fields to fix.

Full description

Check one option algo built from the client's own words with the app's own rules and return it normalised (legs in words, schedule, exits) or the exact fields to fix. Nothing is saved and nothing runs. Ask the client for anything missing instead of choosing it. This is software, not a research analyst or adviser: it builds and checks the client's own rules and never recommends a strategy, a trade or says one will be profitable.

Arguments

algo object, required · The option algo in the format the app's POST /algos takes. Keys: name, legs (1-8), product INTRADAY|OVERNIGHT, entry_time / exit_time "HH:MM" IST, days [1-5] ([] every trading day), entry_mode at_time|wait_and_trade|orb (+ orb {minutes, start, on, break}, wait_pct), sl_pct (per-leg SL %), re_entry, re_entry_max, mtm_target / mtm_stop (INR as text), combined {target / stop: {kind mtm|mtm_pct|delta|theta, value | band, action, times}, lock}, qty_multiplier, monitor (0, 60, 180, 300, 900 s), next_day_exit "HH:MM" (BTST / STBT), underlying, template. A leg is a rule picked at entry time {"rule": {underlying, option_type CE|PE|FUT, strike "ATM" | "OTM 2" | "ITM 1" | {mode: premium|delta|theta|gamma|vega, match, value} | {mode: straddle_width|straddle_pct|pct_atm, value}, expiry weekly|next_weekly|monthly|next_monthly (FUT: current|next), price_basis spot|future|synthetic}, side BUY|SELL, lots} or a fixed contract {instrument (from search_scrip), side, qty, product, order_type MARKET}; any leg may carry exits {target, stop: {kind points|pct|ul_points|ul_pct|delta|theta|gamma|vega, value}, ref, trail, to_cost, on_hit leg|all, re_execute}, wait, orb, lazy {on entry|target|sl, of_leg}, entry (limit / sl_limit). Never send state or a switch: the algo is always saved OFF.
id text, optional · Only to change one of the client's existing algos (it must be switched off); the ids are in get_algos.

`save_algo_off`

Saves OFF

Save an algo switched OFF

Save a new option algo, or change one of the client's algos that is switched off, from the client's own words.

Full description

Save a new option algo, or change one of the client's algos that is switched off, from the client's own words. Always saved switched OFF: it does nothing until the client turns it on himself with its own switch on the Algos screen. You cannot switch it on, start, arm or deploy it, and nothing is placed. Run build_strategy first and show the client what will be saved. This is software, not a research analyst or adviser: it builds and checks the client's own rules and never recommends a strategy, a trade or says one will be profitable.

Arguments

algo object, required · The option algo in the format the app's POST /algos takes. Keys: name, legs (1-8), product INTRADAY|OVERNIGHT, entry_time / exit_time "HH:MM" IST, days [1-5] ([] every trading day), entry_mode at_time|wait_and_trade|orb (+ orb {minutes, start, on, break}, wait_pct), sl_pct (per-leg SL %), re_entry, re_entry_max, mtm_target / mtm_stop (INR as text), combined {target / stop: {kind mtm|mtm_pct|delta|theta, value | band, action, times}, lock}, qty_multiplier, monitor (0, 60, 180, 300, 900 s), next_day_exit "HH:MM" (BTST / STBT), underlying, template. A leg is a rule picked at entry time {"rule": {underlying, option_type CE|PE|FUT, strike "ATM" | "OTM 2" | "ITM 1" | {mode: premium|delta|theta|gamma|vega, match, value} | {mode: straddle_width|straddle_pct|pct_atm, value}, expiry weekly|next_weekly|monthly|next_monthly (FUT: current|next), price_basis spot|future|synthetic}, side BUY|SELL, lots} or a fixed contract {instrument (from search_scrip), side, qty, product, order_type MARKET}; any leg may carry exits {target, stop: {kind points|pct|ul_points|ul_pct|delta|theta|gamma|vega, value}, ref, trail, to_cost, on_hit leg|all, re_execute}, wait, orb, lazy {on entry|target|sl, of_leg}, entry (limit / sl_limit). Never send state or a switch: the algo is always saved OFF.
id text, optional · Only to change one of the client's existing algos (it must be switched off); the ids are in get_algos.

`review_saved_option_strategy`

Checks an algo

Review a change to a saved drawing

Review a complete replacement for one of the client's saved option drawings (Strategy builder). The client reviews and clicks Save in the app; nothing is written or run here. This is software, not a research analyst or adviser: it builds and checks the client's own rules and never recommends a strategy, a trade or says one will be profitable.

Arguments

strategy object, required
- name text, required
- underlying text, required Values: NIFTY, BANKNIFTY, FINNIFTY, MIDCPNIFTY, SENSEX, BANKEX
- expiry text, required
- mul whole number, required
- legs list, required
id text, required · The saved drawing's id.

Act

Prepares an order card. You confirm or reject it in the QuantPower app.

`prepare_order`

Prepares an order

Prepare an order for the client to confirm

Prepare one order and send it to the client's app as a card.

Full description

Prepare one order and send it to the client's app as a card. The CLIENT confirms or rejects it there; you cannot place, confirm, modify or cancel anything. Use search_scrip for the instrument id and get_quote for the price first. Explain in 'reason' why, in the client's words. The card expires in 15 minutes. If the client has two or more brokers connected, 'broker' is required (ask the client which one): without it the call is refused with the list of their brokers by name.

Arguments

order object, required
- instrument text, required · Instrument id from search_scrip, e.g. NSECM:2885 (segment:exchange instrument id).
- side text, required Values: BUY, SELL
- qty whole number, required · Quantity (shares; for F&O, lots x lot size).
- product text, required · MIS intraday, CNC delivery, NRML F&O carry. Values: MIS, CNC, NRML
- order_type text, required Values: MARKET, LIMIT, SL, SL-M
- price text, optional · Limit price in INR as text (LIMIT and SL only).
- trigger text, optional · Trigger price in INR as text (SL and SL-M only).
- validity text, optional, default DAY Values: DAY, IOC, GTD
reason text, required · Why this order, in the client's words, e.g. 'You asked me to add 25 RELIANCE at market'.
broker text, optional · The client's broker connection this order goes to, by name as their broker list shows it, e.g. 'Zerodha' or 'Zerodha (AB••12)'. Needed only when the client has two or more brokers connected; leave it out with one.

`get_prepared_orders`

Reads data

Prepared orders and their status

The orders you prepared for this client and what happened to each: waiting, confirmed (with the order id), rejected, expired or failed. Read-only.

No arguments

Server Wisdom Tree Ventures MCP 0.1.0. Tools behind a QuantPower feature appear when that feature is on for your account.

Plans

Linking Quanty MCP and reading your data through it is free; actions cost tokens at the normal price. See token prices · About Quanty MCP · Quanty inside the app