QuantLogix MCP Server

Connect QuantLogix to your AI tools — Claude Desktop, Claude Code (with plugins), Goose, OpenCode, Google Antigravity, Cursor, or any MCP-aware client. Ask "What's the signal on NVDA?" or "Show me Berkshire's latest 13F" right where you already work.

Add to Claude — one-click OAuth
Add QuantLogix to Claude
The canonical server is https://quantlogix.ai/api/mcp — full catalog, 239 tools, sign in with your QuantLogix account (no ql_ key to paste). Anthropic’s connector directory listing was submitted 2026-06-13 and is awaiting review; until it appears, add QuantLogix as a custom connector.
Add connector in Claude →
https://quantlogix.ai/api/mcp

In Claude: Customize → Connectors → Add custom connector → paste the URL above → sign in. Prefer a headless ql_ key for CI? Use https://quantlogix.ai/api/mcp/v1 (161 tools) — details below.

Free to start — Pro unlocks the advanced tools. One server, one tool catalog, two ways to sign in. Prefer OAuth at https://quantlogix.ai/api/mcp for the full 239-tool catalog. For headless or CI access, the same catalog answers at https://quantlogix.ai/api/mcp/v1 with a ql_... API key from your profile (161 tools — account-scoped Intelligence / Collab / CRM / meetings stay OAuth-only). AI spend is always capped: per key on the key path, per account (a daily cap plus your monthly AI budget) over OAuth. Most market-data and content tools are Free; intraday/options/13F and the screener are Pro.

Endpoint

Canonical (OAuth, 239 tools):

https://quantlogix.ai/api/mcp

API key adapter (161 tools):

https://quantlogix.ai/api/mcp/v1

Protocol version: 2025-06-18 · Transport: streamable-http · Auth: OAuth bearer on /api/mcp, or Authorization: Bearer ql_... on /api/mcp/v1

Setup

Pick your client. OAuth snippets use the canonical URL (no key). Key-path snippets assume QUANTLOGIX_API_KEY in the environment — you can also paste the key inline.

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop:

{
  "mcpServers": {
    "quantlogix": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://quantlogix.ai/api/mcp/v1",
        "--header",
        "Authorization:Bearer ${QUANTLOGIX_API_KEY}"
      ],
      "env": {
        "QUANTLOGIX_API_KEY": "ql_paste_your_key_here"
      }
    }
  }
}

After restart, you'll see quantlogix in the 🔌 connectors menu with the tools listed. Ask "What's QuantLogix's signal for AMD?" to test. Tip: for the full catalog (including your saved Intelligence and QL Collab), add the OAuth connector https://quantlogix.ai/api/mcp via Claude's one-click "Add custom connector" instead — it signs you in, no key needed.

Recommended — install the QuantLogix plugins. You get the MCP tools plus ready-made slash commands (/ql-signal, /ql-screen, /ql-dd, /ql-workflow…), skills, and research subagents. From inside Claude Code:

/plugin marketplace add PortoLogix/quantlogix-plugins
/plugin install quantlogix-signals@quantlogix

Also available: quantlogix-quant (Pro+ agent-builder surface) and quantlogix-research (bull/bear due-diligence). Each plugin bundles the OAuth connector, so the first tool call signs you in — nothing to paste.

Just the tools (no commands)? Add the OAuth connector directly:

claude mcp add --transport http quantlogix https://quantlogix.ai/api/mcp

Then run /mcp and pick Authenticate — a browser opens for the QuantLogix sign-in, and the tools appear once you're back.

Prefer a headless API key (161 tools incl. QL Agents, good for CI)? Add to .mcp.json in the project root:

{
  "mcpServers": {
    "quantlogix": {
      "type": "http",
      "url": "https://quantlogix.ai/api/mcp/v1",
      "headers": {
        "Authorization": "Bearer ql_paste_your_key_here"
      }
    }
  }
}

Mint a ql_... key on your profile. Once connected, the server's built-in prompts (morning_market_brief, pre_earnings_checklist, stock_deep_dive) show up as ready-to-run workflows.

Goose is Block's open-source AI agent. QuantLogix connects as a remote (Streamable HTTP) extension.

Recommended — OAuth connector (no API key). Run:

goose configure

Choose Add Extension → Remote Extension (Streaming HTTP), name it quantlogix, URL https://quantlogix.ai/api/mcp. Goose opens a browser to sign in to QuantLogix, then loads the full catalog. (In Goose Desktop: Settings → Extensions → Add → Remote/Streamable HTTP, same URL.)

Or drop this into ~/.config/goose/config.yaml — the API-key variant needs no browser (mint a ql_... key on your profile):

extensions:
  quantlogix:
    enabled: true
    type: streamable_http
    name: quantlogix
    uri: https://quantlogix.ai/api/mcp/v1
    headers:
      Authorization: "Bearer ql_paste_your_key_here"
    timeout: 300
    description: "QuantLogix — live stock signals, options, 13F & research"

For the full catalog instead, set uri: https://quantlogix.ai/api/mcp and drop the headers block — Goose runs the OAuth sign-in. Test with "What's QuantLogix's signal on NVDA?"

OpenCode is an open-source terminal AI coding agent. It auto-handles OAuth for remote MCP servers, so the full catalog needs just a URL.

Recommended — OAuth connector (no API key). Add to opencode.json (project root or ~/.config/opencode/opencode.json):

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "quantlogix": {
      "type": "remote",
      "url": "https://quantlogix.ai/api/mcp",
      "enabled": true
    }
  }
}

On first tool use OpenCode runs the QuantLogix sign-in in your browser and stores the token — nothing to paste.

Prefer a headless API key (161 tools incl. QL Agents)? Point at /api/mcp/v1 and pass the header (mint a ql_... key on your profile):

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "quantlogix": {
      "type": "remote",
      "url": "https://quantlogix.ai/api/mcp/v1",
      "enabled": true,
      "oauth": false,
      "headers": {
        "Authorization": "Bearer ql_paste_your_key_here"
      }
    }
  }
}

Google Antigravity is Google's agentic IDE. Open its MCP config via the agent panel … → MCP Servers → Manage → View raw config (or Settings → Customizations → Open MCP Config) — the file lives at ~/.gemini/config/mcp_config.json (global) or .agents/mcp_config.json (per-workspace).

Note: Antigravity uses serverUrl (not url) for remote servers. Add a ql_... key from your profile:

{
  "mcpServers": {
    "quantlogix": {
      "serverUrl": "https://quantlogix.ai/api/mcp/v1",
      "headers": {
        "Authorization": "Bearer ql_paste_your_key_here"
      }
    }
  }
}

That is the 161-tool API-key catalog (everything except your account-scoped Intelligence, Collab, CRM and meetings). If your Antigravity build supports MCP OAuth, point serverUrl at https://quantlogix.ai/api/mcp and drop the headers block for the full 239-tool catalog with account sign-in. New MCP tools run in Ask mode until you allow them (e.g. mcp(quantlogix/*)).

Cursor has no plugin marketplace, but it's a full MCP client. Two files give you parity — the connection and a rule that teaches the agent how to use the tools.

1. Connect — .cursor/mcp.json (project) or ~/.cursor/mcp.json (global). OAuth connector, no key on recent Cursor:

{
  "mcpServers": {
    "quantlogix": { "url": "https://quantlogix.ai/api/mcp" }
  }
}

Enable it under Settings → MCP. Older Cursor (no MCP OAuth): use https://quantlogix.ai/api/mcp/v1 with an Authorization: Bearer ql_... header (key from your profile).

2. Add the rule — .cursor/rules/quantlogix.mdc. It routes stock/market questions through the QuantLogix tools (which tool for what, no invented numbers, disclaimers) — the Cursor-native equivalent of our Claude Code plugin skill. Download quantlogix.mdc and drop it into .cursor/rules/.

List available tools:

curl -s https://quantlogix.ai/api/mcp/v1 \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Call a tool (requires a ql_... key — Free tier works for signal):

curl -s https://quantlogix.ai/api/mcp/v1 \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ql_paste_your_key_here" \
  -d '{
    "jsonrpc":"2.0",
    "id":2,
    "method":"tools/call",
    "params":{
      "name":"signal",
      "arguments":{"ticker":"NVDA"}
    }
  }'

The ql CLI

Everything on the OAuth connector is also available from your terminal — CRM (contacts, deals, tasks, goals & metrics, live report), QL Notes, Sheets, Slides, Draw diagrams, and every market-data tool. Zero dependencies; needs Node 18+.

# install
curl -fsSL https://quantlogix.ai/cli/ql.mjs -o /usr/local/bin/ql && chmod +x /usr/local/bin/ql

# one-time browser sign-in (OAuth + PKCE — same auth as the connector)
ql login

# try it
ql signal NVDA                         # 5-factor signal (free)
ql day NVDA                            # QDTSS intraday signal (Pro+)
ql screener sector=Technology          # the Pro screener (Starter+)
ql smart-money                         # Congress + 13F consensus (Starter+)
ql private financials stripe           # dilution + public-comps bridge (Pro+)
ql marks confidence=high               # QL Private Marks board (Institutional)
ql simulate NVDA:50000,TLT:30000       # Monte Carlo what-if book (Institutional)
ql webhooks add url=https://ci.example # outbound event push (Pro+)
ql ask "is NVDA overextended?"         # Ask QuantLogix (Pro+, metered)
ql crm report                          # live pipeline + goal pace
ql notes new title="Meeting notes" --file notes.md
ql backtest SPY trend                  # strategy vs buy-and-hold (Starter+)
ql portfolio NVDA:50,AAPL:30           # 5-factor book roll-up (Pro+)
ql overlay NVDA:50,AAPL:30             # Portfolio Alpha overlay (every tier)
ql call get_signal ticker=NVDA         # anything else in the catalog
ql help                                # the full command map

Available Tools

The catalog below is the shared tool set — 239 tools — served by both /api/mcp (OAuth, all 239) and /api/mcp/v1 (API key, 161 of them), and by the ql CLI. Each tier badge is the minimum access required; every tool inherits the same gate as its underlying QuantLogix endpoint. Tools marked sign-in required are OAuth-only; everything else — including ask and run_agent — works through either credential. Older tool names still work: the key server's original names (signal, agents_list, day_signal, …) remain permanent aliases for their verb_noun equivalents (get_signal, list_agents, get_day_signal), so existing configs and saved prompts keep running unchanged.

Free any signed-in account Pro Pro / Institutional Tiered monthly cap scales by subscription Write creates / edits data

📈 Market Data & Signals Free

get_signalLong-term 5-factor composite signal — Bullish/Neutral/Bearish, score 0-100, per-factor breakdown (technical, momentum, fundamental, options, microstructure), pattern prediction. (get_long_term_signal is the same engine.)
get_quoteLive price quote for a stock or ETF.
get_signal_rankingsTop-N tickers ranked by composite signal score, optionally filtered.
compare_tickersHead-to-head 5-factor comparison of 2–8 tickers — ranked by composite score with each name's per-factor breakdown, plus a verdict naming the strongest and the factor it leads on. Doubles as watchlist triage.
get_market_overviewIndices, VIX, and broad-market snapshot.
get_market_trendsDaily Trend Monitor — quantitative trend read per major index (score, direction, flags, key MA levels), SPY regime with flip probability, private-markets WoW trend, and detected trend flips.
get_macroUS macro indicators (FRED) — rates, inflation, employment, growth, with a composite macro-sentiment score.
get_track_recordVerified signal track record — win-rate, returns, and the four validation grades behind /proof.
get_ipo_calendarUpcoming + recent IPOs.
get_price_chartRendered price-chart image for a ticker.
get_heatmapS&P 500 sector performance heatmap — the 11 GICS sectors colored green→red by today's move, as an inline image (+ leaders/laggards text). No ticker needed.

📰 QuantLogix Content Free

list_wires · get_wireQL Wire — short AI-authored market wires tied to S&P 500 tickers (analyst-call recaps, earnings reactions, sector moves). List + full body.
list_ql_updates · get_ql_updateQL Updates — QuantLogix editorial / product posts (VC intelligence, macro strategy, persona briefs). List + full body.
list_briefings · get_briefingDaily / weekly / quarterly market briefings. List (kind filter) + full body.

🎯 QL Oracle — Verified Data Free

The verified stack as a data source: records graded by the platform against real prices — never self-reported. Every payload carries an attribution block (source, methodology, disclaimer, cite_as) so agents cite QuantLogix when using the data. Same shapes as GET /api/v1/verified.

get_verified_convictionThe Conviction Book — raw crowd vs track-record-weighted consensus per ticker, plus the sharpest crowd-vs-verified divergences. Optional ticker filter.
get_verified_indexThe QL Verified Index — the live long/short model book, cumulative level (inception 100), and equity curve.
get_verified_creatorsThe Creator Leaderboard — approved analysts ranked by the graded win rate of the agents they publish.
get_verified_recordOne creator's aggregate verified record by handle — the portable answer to "has this analyst actually been right?".

🕵️ Market Intelligence Free

Who else is positioned, how crowded the trade is, and what is scheduled next — the disclosure, positioning and statistical layers behind the signal.

get_congress_tradesUS Congress STOCK Act disclosures. Per-ticker card (trade count, distinct politicians, buy/sell split, net direction, party split, estimated volume) or a multi-politician cluster leaderboard. Filings lag the trade by up to 45 days by law.
get_insider_activitySEC Form 4 insider transactions from EDGAR — buy/sell counts, net direction, a Bullish/Neutral/Bearish read, and whether the filings form a cluster (several insiders acting together, the configuration that carries weight).
get_short_interest · scan_squeezeShort interest, days-to-cover, % of float and the daily short-volume stream for one ticker — or the ranked market-wide squeeze board scored on SI/float, days-to-cover, float size and momentum ignition.
get_sector_rotationRelative Rotation Graph for the 11 SPDR sectors vs SPY — RS-Ratio, RS-Momentum, quadrant (Leading/Weakening/Lagging/Improving) and a 10-week rotation tail per sector.
get_market_outliersThe day's universe-wide statistical outliers — moves and volume extreme against the cross-section (z-score + relative-volume thresholds over ~5,000 names), not merely large. Filter up / down / volume.
get_catalystsForward catalyst tape — earnings, IPOs, dividends, splits, macro prints, opex and disclosed Congress trades, market-wide or scoped to a ticker. Tier-laddered and clamped rather than refused: free 7 days (earnings/macro/opex), Starter 21 days all types, Pro+ the full 60-day tape. The response echoes tier, limits and clamped.
get_financialsReported income statement, balance sheet and cash flow from SEC filings — quarterly or annual, newest first, with fiscal period and filing date.
get_earnings_drift · get_dip_stats · get_vol_profileThe statistical bases behind the free tools: how a name historically trades earnings (beat rate, median absolute reaction, drift), where it sits against its own drawdown history, and its realized-vol regime with expected-move bands.
get_promise_ledgerManagement Promise Ledger — every forward commitment from earnings-call transcripts stamped kept / missed / walked-back / pending, with the exact triggering sentence as the receipt, plus a 0–100 credibility index and archetype. Pass demo=true (or ticker MRDC) for the fictional Meridian Compute ledger — no transcript quota. Both doors, Free.

🧺 Strategy Baskets Free to browse Tiered to build

Browsing the curated catalog and the theme taxonomy is free for everyone. Resolving a theme into a live basket is metered by subscription tier — so basket-building availability is tier-based, not unlimited on Free.

list_baskets · get_basket FreeCurated strategy/thematic baskets catalog + detail (constituents, live track record).
list_basket_themes FreeBrowse the 487-theme investment taxonomy.
resolve_theme_basket Write TieredResolve a theme into a live equal-weight basket. Monthly distinct-theme cap by tier — Free 5 · Starter 25 · Pro 100 · Institutional unlimited. The cap is shared with the website, so switching surfaces doesn't reset it; admins bypass.

🔬 Advanced Signals & Fundamentals Pro

get_day_signalIntraday QDTSS signal — 4-layer scoring with VWAP, ORB, session classifier, trade plan.
get_swing_signalSwing signal — A/B/C/D conviction grade + 3-pillar breakdown (Technical / Flow / Sentiment), 2d-4wk.
get_microstructureOrder-flow microstructure — VPIN, OFI, smart-money score, institutional bias, volume profile, spread.
get_options_flowUnusual options activity — large/sweeps, premium, sentiment.
get_options_dataFull Options-tab data — pick an action: chain (Greeks + IV + max-pain + put/call), iv_skew (surface + term structure), large_flow, gex (gamma exposure, dealer flip, call/put walls), earnings_move (ATM-straddle implied move), expirations.
get_options_strategyRanked multi-leg options strategies for a thesis (bullish/bearish/neutral/volatility/earnings × short/medium/long) — concrete strikes + expirations from the live chain, debit/credit, max P/L, breakevens, delta-POP, IV-rank context.
get_earnings_calendar · get_earnings_setupPer-quarter EPS/revenue actuals + estimates, and the combined earnings-setup read (implied move + engine signal + revision momentum).
get_corporate_eventsDividends, splits, spinoffs, M&A for a ticker.
get_institutional_holders · get_institutional_fundTop 13F holders for a ticker (Berkshire, BlackRock, Vanguard, Renaissance, Citadel, …) and a fund's full 13F portfolio by CIK.
list_private_companies · get_private_company · get_private_company_financials~30 top private / pre-IPO companies (Anthropic, OpenAI, Stripe, Databricks, Anduril, …) + single-company profile + dilution waterfall / financing scorecard / public-comp valuation bridge. Profiles carry the live enrichment overlay: freshest sourced valuation mark + live-tracked C-suite leadership changes.
get_shadow_bookQL Shadow Book — ranked private-market shadow of a public book (holdings, watchlist, theses, or an explicit tickers list). Book posture, IPO-window names, deal ripples, investor slice. Deterministic; no model call. Pass demo=true for NVDA/MSFT/GOOGL. Both doors, Pro+.
get_readthrough_basketPrivate→public read-through basket — generated equal-weight sleeve from the crossmarket graph (named rivals first, then suppliers/investors, then customers/comps). Live 5-factor roll-up. Holdings rebuild when the graph changes. Pass demo=true (or slug helios-robotics) for the fictional Helios Robotics sleeve. Both doors, Pro+.
get_ipo_readiness · get_private_liquidity · get_investor_fitPrivate-market intelligence engines: 5-axis IPO-readiness composite (single company or ranked universe), secondary-market liquidity read (marks, instruments, restriction/timing model), and "who would invest right now" investor-fit ranking (deployment pace, dry powder, check-size/sector/stage fit).
list_investors · get_investor~280 VC / PE / CVC / Sovereign-wealth firms (Andreessen Horowitz, Sequoia, Founders Fund, Thrive, Tiger Global, …) — AUM, active fund, stage + sector focus, notable partners + portfolio, recent investments. Directory + single-firm profile.
screenerMulti-criteria Pro Screener across stocks + ETFs. Pro tier only on the API-key path.
simulate_portfolioQL Simulator (Institutional): roll a dollar-weighted what-if book through thousands of Monte Carlo futures bootstrapped from real daily history — optionally regime-switching (a vol/trend Markov chain learned from the book itself; "stressed" starts in today's stressed twin). Returns the forward distribution with VaR/CVaR, the regime context (dwell, shift risk, regime drift), a model-challenge read against the textbook lognormal and a zero-drift challenger, and a validation-first confidence tag. Stateless.
get_index_radarIndex Reconstitution Radar (Institutional): who's next INTO the S&P 500 and who's drifting toward the exit — every non-member scored against the published eligibility criteria (unadjusted-cap floor, TTM GAAP earnings, 12-month seasoning) with deterministic readiness scores and per-component parts, the mega-IPO seasoning clock with exact eligible-on dates, the deletion-watch tail, structurally-ineligible giants (foreign domicile / LPs), day-over-day newly-qualified diffs, and the forced passive bid an addition would trigger at your tracked-AUM assumption. Same engine as /index-radar.
get_hedging_deskQL Hedging Desk (Institutional): the dealer hedging response surface, not a static gamma snapshot — how much stock market makers must mechanically buy or sell to stay delta-flat across a joint spot × vol × time grid. Adds the two channels a gamma chart structurally cannot show: vanna (the vol-spot feedback loop that turns a vol bid into forced selling) and charm (the decay pin that arrives on the calendar). Every scenario is produced by fully repricing the dealer book at the shifted state rather than multiplying today's greeks by a move, so it captures the convexity a greeks-times-move estimate misses. Returns the response grid, a spot ladder flagging where hedging amplifies vs dampens, isolated vol/time ladders, six named scenarios, the gamma- and vanna-flip prices, the accelerant band around spot, per-strike decomposition, and — because dealer positioning is assumed from open interest, never observed — the same headline numbers under the opposite sign convention with a warning when the two disagree. Flows in shares, dollars and % of 20-session ADV. Same engine as /hedging-desk.
get_whale_tapeQL Whale Tape: real large options orders from the OPRA trade tape (15-minute delayed on our data tier), graded in public. Prints are grouped into orders; execution class comes from exchange sale conditions (SWEEP = Intermarket Sweep flag, SPLIT = multi-venue fill, FLOOR / CROSS / AUCTION / PRINT, MULTI-LEG never graded); the side is a tick test against the prior print and every order states its sentiment_basis because option bid/ask is not in the tier. Returns the last five sessions of orders (largest or latest, filter by ticker / side / premium), tiers, repeat-whale counts, and the forward-only record — one $250k+ single-leg order per ticker × day × side graded by the shared expiry rule, sliced by execution, basis, moneyness, DTE, tier and repeat. Free for every tier. Same data as /optionsflow-pro.
get_large_positioningQL Large Positioning: window-aggregate options positioning from chain snapshots, graded in public — the complement to get_whale_tape. Every 30 minutes the most active chains are snapshotted near-the-money and the increment of volume × price per contract is the premium that accumulated in the window (events ≥ $250k and 100 contracts; not individual prints). Direction is the contract side; opening vs closing is classified next session from open interest. Returns the last five sessions of events and the forward-only record (one call per ticker × day × side, hit rate / expectancy by side, open-vs-close, moneyness). Free for every tier.
get_ipo_ledger FreeQL IPO Ledger: QuantLogix's dated pre-IPO valuation calls — a range published BEFORE a deal prices and graded in public when the company lists (inside the range / above it / below it). Returns the scoreboard (hit rate, per-stage and sharp-call breakouts, median band width), every open call with its published range and band width, and the graded record; optionally the rumored watchlist (indicative ranges, graded nowhere) and the stress lab (expected hit rate across an IPO-window repricing grid — a calibration check on our own bands, never a forecast). Ranges are derived from the last confirmed private mark times a published band per filing stage, never hand-typed per name. Forward only: a listing that predates its call is voided, never counted, and the hit rate stays null until something actually grades. Same engine as /ipo-ledger.
get_regime_history FreeAlpha Clusters regime intelligence: how the twelve Street-vs-Engine regimes behave over time, not just today's snapshot. Returns the transition matrix between the previous trading day's labels and the latest (with the individual moves worth reading), persistence per regime (hold rate + median run length), and the REALIZED forward return of the names each regime labelled — credited to the regime held at the START of the interval and never relabelled afterwards. A regime with fewer than three observations reports nothing rather than an average built on one name; day one is an honest cold start with no transitions. Same engine as /alpha-clusters.
get_ql_record FreeQL Record: QuantLogix's consolidated, public, hash-chained track record — every call the platform grades (engine-vs-Street, pre-IPO valuation ranges at listing, directional regime labels) in one scoreboard with the date each was made and the date the market graded it. Misses listed exactly as prominently as hits; hit rate excludes pushes and is null until something has actually graded. Aggregate disciplines (the signal walk-forward, including its verdict when unflattering) sit beside the record and never enter the chained rate. Every entry is SHA-256 chained and the head hash is returned for verification. Same engine as /record.
list_strategy_backtests FreeCatalog of runnable strategy-backtest IDs (buy_hold, trend, golden_cross, rsi_meanrev, macd) + descriptions + the full metrics contract every backtest returns. Call this before run_strategy_backtest so an assistant can pick a valid strategy id.
run_strategy_backtest Starter+Point-in-time historical replay of a classic timing strategy on ONE instrument (ticker or ETF) vs buy-and-hold — the same engine the Alpha Engine Week-10 Strategy Backtest Lab uses. Returns strategy vs B&H metrics (total return, CAGR, Sharpe, max drawdown, exposure, trade count), the per-metric edge delta, and a walk-forward (train 70% / test 30%) Sharpe pair with an OVERFIT / MILD_DECAY / ROBUST / NO_EDGE verdict. Years 1–10 (default 10). Pure math over daily closes — no AI tokens billed. Distinct from the in-app agent tool run_backtest (indicator-DSL presets). Hypothetical / educational.
analyze_portfolioRoll the 5-factor engine up over a whole portfolio or watchlist (up to 25 holdings, weighted by percent / market value / shares). Returns the weighted composite posture, signal mix by weight, weighted per-factor exposure, position + sector concentration (HHI, top-position, effective #positions), plain-English risk flags, and a prioritized list of the positions most worth reviewing.
get_portfolio_analyticsPortfolio Desk analytics — the institutional replay of a book vs SPY over up to 3 years: CAGR, Sharpe / Sortino / Calmar, drawdown with recovery, beta / alpha / R² / tracking error / information ratio, capture, VaR / CVaR, skew / kurtosis, Euler risk decomposition, contributions, asset mix, monthly table, growth + underwater series, notes and projection inputs. Pass positions="NVDA:10,MSFT:$5000,AAPL" for a stateless replay on either door; OAuth sessions may omit it to read the signed-in book with the household balance sheet and the realized daily-snapshot return. Labelled a backcast, never a track record. REST twin: POST /api/v1/portfolio/analytics. Both doors, Pro+.
check_portfolio_healthFive-dimension health check on a set of holdings — concentration, diversification, signal quality, risk positions and sector exposure — rolled into a 0-100 score and letter grade, with the graded per-position table. Grades structural risk, where analyze_portfolio gives posture and portfolio_alpha_overlay gives the action queue.
portfolio_alpha_overlay Free to run Pro for your live bookThe Portfolio Alpha overlay — what to do, not just what the book scores. Every position gets a verdict (agree_hold / ql_says_sell / ql_says_add) and an action (CLOSE / ADD / HOLD / WATCH) with its reason, the 5-factor breakdown and the engine's historical accuracy on that name. On top: action_items, one priority-sorted queue (CLOSE → ADD → DIVERSIFY → HEDGE); signal_gaps, strong-buy names the book does not hold; today's market regime; and an A–F risk grade with its concentration + signal drivers. Pass holdings to overlay any book on either credential — free tier up to 8 holdings, Starter+ up to 25. Omit holdings to run against your own connected brokerage positions instead: that reads live account data, so it needs Pro+ and the OAuth connector.

🧠 Your Saved Intelligence Free

Your own saved AI research / chat sessions. Sign-in required (OAuth connector) — every call is scoped to your account; you can never read another user's data.

list_intelligence_articles · get_intelligence_article · search_intelligenceList, read, and full-text-search your saved QL Intelligence articles (research / due-diligence / Monte-Carlo / saved chat sessions).

⚙️ Your QL Agents Free

Set up + manage your own scheduled AI agents (a recurring digest, a conditional alert, or a one-time scheduled run). Sign-in required (OAuth connector); every call is scoped to your account. Create is capped per tier (Free 1 / Starter 3 / Pro 15 / Institutional 50). Agents run on the server schedule and notify you by push + email — manage them anytime at /agents.

list_agents · get_agentList your agents (name, mode, schedule, enabled state, run count) and read one in detail with its recent run history + performance stats (success rate, fired rate, errors, spend).
create_agent · update_agent WriteCreate a digest/alert agent on a premarket/close/weekly schedule — or a one-time run at a specific date & time (cadence once + run_at, fires once then auto-pauses). Update reschedules / edits / enables / pauses; deletion stays on the web at /agents.

👥 Your QL Collab Workspace Free

Your team's Collab spaces (Clerk-org-scoped). Reads + writes are restricted to spaces you belong to; a write re-checks membership on the row first.

list_collab_spaces · list_collab_items · get_collab_item · search_collabBrowse your team channels and read/search their messages, notes, sheets, slides, and diagrams.
create_collab_note · update_collab_note WriteCreate / edit a Notion-style markdown note in a Collab space.
create_collab_sheet · update_collab_sheet WriteCreate / edit a live spreadsheet (formulas + live market-data functions; exports to CSV/Excel).
read_sheet_range · write_sheet_range · get_note_section · append_note_section WriteDocuments as APIs — read or write ONE A1 range of a QL Sheet or ONE heading section of a QL Note; everything outside the addressed piece is untouched. REST twin: /api/v1/collab (ql_ key, Pro+). sign-in required
post_collab_message · log_collab_decision · get_collab_record WriteChannels as APIs — post into a team channel as the caller (every $TICKER point-in-time pinned on arrival), log a tracked call the resolver grades, and read the channel Record ("what we said vs what happened"). Catalog + competitor matrix: /collab-tools. sign-in required
create_collab_diagram · update_collab_diagram WriteGenerate / edit a Mermaid diagram (flowchart, sequence, entity graph, mindmap, gantt).
create_collab_deck · update_collab_deck WriteCreate / edit a Marp markdown slide deck (presentable, exports to PDF).

📇 Your CRM Starter+

Your /crm — goal & metrics driven. Sign-in required (OAuth connector); every call is scoped to your contacts, deals, tasks, and goals. Same Starter+ gate as the app.

get_crm_reportLive analytics rollup — contact pipeline, deal pipeline value (open / weighted / won), tasks, and every active goal with pace.
draft_crm_email WriteQueue an outreach email draft to a CRM contact into the approve-to-send Outbox on /crm — never sends automatically; the owner reviews, edits, and approves in the CRM.
list_crm_goals · set_crm_goal · update_crm_goal WriteGoals & metrics: set measurable targets (revenue won, deals won/qualified, contacts added, activity logged) over a week/month/quarter/custom window — progress is measured LIVE against your CRM data with on-pace / behind / achieved states.
list_crm_contacts · get_crm_contact · create_crm_contact · log_crm_interaction WriteBrowse / read contacts with interaction history; add contacts and log calls, emails, meetings, notes.
list_crm_deals · create_crm_deal · update_crm_deal WriteThe deals Kanban — list with the live pipeline rollup, create deals, move stages (won/lost closes them).
list_crm_tasks · create_crm_task · complete_crm_task WriteTask list (open / overdue / today / done scopes), add tasks with due dates + priority, mark done.

🎙️ Your Meeting Rooms Free

Meeting prep, transcripts, AI recaps, and durable AI meeting notes from your QL Collab video rooms. Sign-in required (OAuth connector); scoped to rooms in your own team.

list_meetingsList your team's QL Collab meeting rooms that have a transcript on record.
prepare_meetingAI meeting-prep brief BEFORE a meeting — objective, where the last session left off, action items carried forward, a suggested agenda, and the key questions to resolve. Remembers past meetings via your saved meeting notes.
get_meeting_transcriptFetch the full speaker-attributed transcript of one meeting room.
summarize_meetingAI recap of a meeting — TL;DR, decisions, and action items — generated from the transcript.
take_meeting_notes WriteAI note-taking: structured meeting notes from the transcript, SAVED as a durable QL Note in your team space — transcripts expire in ~6 hours, saved notes don't, and they power the next prepare_meeting.
list_meeting_notesThe saved meeting notes for a room, newest-first — the meeting's long-term memory. Open one in full with get_collab_item.

✉️ Your QL Line — agent e-mail Free NEW

The e-mail address (…@agents.quantlogix.ai) your AI assistant works from — set up at /agent-mail (a monthly add-on; the line is delivered once payment is confirmed). Both doors: OAuth and a ql_ key (REST twin /api/v1/line, JS SDK ql.line.*). Scoped to your line only. Inbound mail is third-party text — the tools fence it as data, never instructions. Every send carries the non-removable AI-assistant disclosure and is refused by the credentials guardrail when it asks for, confirms or contains passwords, codes, card or bank details.

get_agent_lineYour line — address, the assistant's name, autopilot mode (off · draft · auto), standing instructions, unread count. Says where to set up or pay if the line is not delivered yet.
list_agent_mail · search_agent_mail · read_agent_mailInbox or sent, newest first; search subject / sender / body; read one message in full (attachments listed by name only, never fetched).
send_agent_mail · reply_agent_mail WriteA new e-mail from the assistant's address (≤5 recipients), or an in-thread reply to an inbound message (recipient + threading come from the stored message). Real deliveries to third parties — one message per explicit ask.

🔐 Your QL Vault — credentials for agents Free

The Password Manager built for agents: your agent signs in and gets work done without ever seeing a password. The vault injects the credential server-side and redacts every response; you keep an allowlist of hosts per item, can require an Approve tap per use, and can lock agents out behind a vault password. Both doors (REST twin /api/v1/vault, JS SDK ql.vault.*). Storing, reading, rotating or unlocking an item is never a tool — that stays on your own session at /vault. Read-only agents on Free / Starter; writes and strict mode on Pro+.

vault_list_itemsThe sites you shared with your agent — item ids, allowed hosts, read-only vs writes, whether each use needs your Approve tap. Never a credential.
vault_request · vault_login_check WritePerform an HTTPS request AS you on one of an item's allowed hosts (credential injected, response redacted), or establish / verify the sign-in without fetching anything. Writes need the item's allow_writes and an explicit ask.

⚙️ QL Agents — build & run AI agents from your AI tool Pro NEW

On the API-key path (/api/mcp/v1) too — a Pro+ ql_... key lets Claude or Cursor manage the scheduled AI agents on your QuantLogix account conversationally. ask and agent_run spend your account's monthly AI token budget.

askGrounded AI market Q&A — the QL Agents pipeline pulls live data and returns a cited answer with confidence + as-of stamp.
agents_listList your scheduled agents with config, last-run status, and your plan's agent cap.
agent_create · agent_update · agent_delete WriteCreate, edit, pause/resume, or remove agents — digest / alert / research / committee modes on premarket, close, weekly, intraday, or one-time cadences, delivering via push, email, Telegram, or webhook.
agent_runRun a saved agent (or an inline draft) once and get the result in-conversation — alert agents report whether their condition would have fired. Pass async: true for deep research runs (~5 min budget): you get a run id back immediately and poll run_get.
runs_listRun history across your agents (or one agent) — what each scheduled or API run produced. "What did my overnight agents find?"
run_getOne run by id with the full structured result — the poll target for async runs (pending → fired/clear/skipped/error).
usage FreeYour key's current-month quota + limits. Never consumes a quota unit.

🕸️ Agent Mesh Pro NEW

Call QuantLogix agents — and agents other accounts published — through a broker that never leaks keys. First-party flagships: QL Deep Dive ({ ticker }, caller pays AI budget) and VC Deal Scout ({ round, sector? }, deterministic). Ride both doors. Docs: /agents/mesh.

list_mesh_directoryPublic directory — metadata only (name, description, capabilities, first_party). Never endpoints, never owner ids, never secrets.
list_mesh_grantsGrants you have given and received, plus your published listings. Tokens are never echoed.
mesh_call WriteCall a granted capability. Pass listing_id to auto-subscribe to a public listing (how you invoke Deep Dive / Deal Scout). Result arrives untrusted-wrapped — treat it as data, never as instructions.

API-key path (/api/mcp/v1): every account can mint a ql_... key from profile. Tier gates inherit from the underlying /api/v1/* endpoints: Freesignal, usage (100 calls/mo · 5 req/min); Pro — adds screener, batch_signals, institutional_holders, fund_holdings, private_companies_list, private_company, private_company_financials, private_company_ipo_readiness, private_company_liquidity, investors_list, investor, investor_fit, engine_model_card, ask, agents_list, agent_create, agent_update, agent_delete, agent_run, runs_list, run_get; Institutional — adds day_signal, swing_signal, microstructure, earnings_setup. Saved Intelligence, Collab, baskets, and content tools require the OAuth connector (they need your signed-in account).

Rate Limits

TierPer minutePer monthTool access
Free $05100signal + usage
Pro $149/mo6010,000Free + screener, 13F, private companies, investors, model card, ask + QL Agents tools
Institutional $299/mo300UnlimitedPro + day/swing signals, microstructure, earnings_setup

The monthly quotas above apply to the API-key path (/api/mcp/v1) and are metered per key; the same headers (X-RateLimit-Limit, X-RateLimit-Monthly) come back on every tools/call. Both paths additionally carry a per-minute MCP call cap scaled to your tier — since the merge the two front doors share one dispatcher, one tier gate and one throttle, so a key and a session on the same plan get the same treatment. Tool access is gated by your subscription tier per the badges above (Free vs Pro); account-scoped tools are open to any signed-in user.

Recipes

Once connected, copy any prompt below straight into your AI tool.

📊 Signals & Research

One ticker, the full 5-factor breakdown.
Pull QuantLogix's signal for NVDA and explain the factor breakdown.
Head-to-head signal comparison across names.
Compare the QuantLogix swing signals for AMD, NVDA, and AVGO.
Intraday signal with an actionable trade plan.
Get QuantLogix's intraday day-signal for SPY and tell me the trade plan.
Freeform market question, answered with live data + citations.Pro
Ask QuantLogix whether NVDA looks extended after this run.
Who's holding a name, per the latest 13F.
Who are the top institutional holders of TSLA according to QuantLogix?
Private-market names plus their latest headlines.
List private companies in the AI sector and show me OpenAI's latest financing round.

💼 Portfolio & Risk

Roll the 5-factor engine up over a whole portfolio.Pro
Here's my portfolio — NVDA 40%, AAPL 30%, KO 30%. Run QuantLogix's 5-factor analysis and tell me my concentration risk and which positions to review.

⚙️ Agents & Automation

Stand up a recurring digest agent.Pro
Create a QL agent that watches my chip names premarket and emails me a digest.
Run a saved agent on demand.Pro
Run my Portfolio Guardian agent now and show me what it says.

Get Started

You'll need an active Pro or Institutional subscription, then a key from the API Keys card on your profile.

Get an API Key View Pricing REST API Docs

Protocol Notes

The QuantLogix MCP server implements:

Tool calls require a credential: Authorization: Bearer <oauth-token> on /api/mcp, or Authorization: Bearer ql_... on /api/mcp/v1. The initialize, ping, and tools/list methods are public so clients can probe the server before authenticating. Upstream errors (rate limit, invalid ticker, tier gate) come back as MCP isError content so the AI can recover and explain to the user in-conversation, rather than crashing the client. A tool called through the wrong door returns JSON-RPC -32003 naming the credential it needs.

Note: passing the key in the query string (?apikey=) is no longer accepted on /api/mcp/v1 — query strings are written verbatim into server and CDN access logs, which turns the key into a logged credential. Use the Authorization or X-API-Key header.

Frequently asked questions

What is the QuantLogix MCP server? One MCP registry: 239 tools through OAuth at /api/mcp and a 161-tool API-key subset at /api/mcp/v1. Connect Claude Desktop, Claude Code, Cursor, Goose, OpenCode, or any MCP-aware client to live signals, private-company research, and agent workflows.
Should I use OAuth or an API key? Use OAuth at /api/mcp for one-click connector sign-in and the full 239-tool registry. Use a ql_ key at /api/mcp/v1 when your client only supports bearer-token auth (161 tools; account-scoped Intelligence, Collab, CRM and meetings remain OAuth-only). The same keys work with the REST API at /api-docs.
Which tools are free vs Pro? Free credentials cover core signal and news tools. Pro and Institutional unlock screener, 13F, private companies, quant-lab tools, and the engine model card. tools/list only returns tools your credential can call.
How do I authenticate tool calls? Send Authorization: Bearer <oauth-token> on /api/mcp, or Authorization: Bearer ql_... / X-API-Key on /api/mcp/v1. initialize, ping, and tools/list are public so clients can probe before sign-in.

Questions or feedback? Contact our team.