Bloomberg MCP
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Bloomberg MCPShow me the top holders for AAPL"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Bloomberg MCP
A Model Context Protocol server that gives AI assistants direct access to Bloomberg Terminal data.
Bloomberg MCP bridges the Bloomberg Terminal with AI assistants via the Model Context Protocol. It exposes 18 tools covering reference data, bulk data, historical analysis, technical analysis, estimates, ownership, supply chain, screening, BQL queries, and calendars — all accessible through natural language.
Enhanced fork of tallinn102/bloomberg-mcp — v1.1 adds 6 new tools, modular architecture, caching layer, and 10 analytical FieldSets.
You: "What are the top holders and supply chain for CEG US Equity?"
Claude: runs bloomberg_get_bulk_data with TOP_20_HOLDERS_PUBLIC_FILINGS
runs bloomberg_get_supply_chain with suppliers + customers
→ Returns structured holder list + supplier/customer networkArchitecture
graph TB
subgraph Clients
CC[Claude Code]
WC[Web Client]
CA[Custom App]
end
subgraph "Bloomberg MCP Server"
direction TB
MCP["FastMCP Server<br/><i>18 tools exposed</i>"]
subgraph Handlers["Handler Layer"]
direction LR
REF[Reference & Historical]
BULK[Bulk Data & Estimates]
TA[Technical Analysis]
SCREEN[Screening & BQL]
OWN[Ownership & Supply Chain]
CAL[Calendars]
end
subgraph Core["Core Layer"]
direction LR
SESSION["BloombergSession<br/><i>Singleton + Cache</i>"]
REQ[Request Builder]
RESP[Response Parser]
end
end
BBG["Bloomberg Terminal<br/><i>blpapi on port 8194</i>"]
CC -- stdio --> MCP
WC -- HTTP/SSE --> MCP
CA -- HTTP/SSE --> MCP
MCP --> Handlers
Handlers --> Core
SESSION <--> BBG
REQ --> SESSION
SESSION --> RESP
style MCP fill:#1a73e8,stroke:#1557b0,color:#fff
style BBG fill:#ff6f00,stroke:#e65100,color:#fff
style SESSION fill:#2e7d32,stroke:#1b5e20,color:#fffRelated MCP server: FinClaw
Tools Overview (18 tools)
graph LR
subgraph "Market Data (4)"
T1["bloomberg_get_reference_data<br/><i>BDP snapshots</i>"]
T2["bloomberg_get_historical_data<br/><i>BDH time series</i>"]
T3["bloomberg_get_intraday_bars<br/><i>OHLCV candles</i>"]
T4["bloomberg_get_intraday_ticks<br/><i>Raw ticks</i>"]
end
subgraph "Bulk Data & Analysis (4)"
T5["bloomberg_get_bulk_data<br/><i>BDS tables</i>"]
T6["bloomberg_get_estimates_detail<br/><i>Multi-period consensus</i>"]
T7["bloomberg_get_ownership<br/><i>Holder analysis</i>"]
T8["bloomberg_get_supply_chain<br/><i>SPLC network</i>"]
end
subgraph "Technical & BQL (2)"
T9["bloomberg_get_technical_analysis<br/><i>RSI, MACD, Bollinger...</i>"]
T10["bloomberg_run_bql<br/><i>Query language</i>"]
end
subgraph "Discovery (3)"
T11["bloomberg_search_securities<br/><i>Find by name/ticker</i>"]
T12["bloomberg_search_fields<br/><i>Field mnemonics</i>"]
T13["bloomberg_get_field_info<br/><i>Field metadata</i>"]
end
subgraph "Screening (3)"
T14["bloomberg_run_screen<br/><i>Saved EQS screens</i>"]
T15["bloomberg_get_universe<br/><i>Index constituents</i>"]
T16["bloomberg_dynamic_screen<br/><i>Custom filter + rank</i>"]
end
subgraph "Calendars (2)"
T17["bloomberg_get_economic_calendar<br/><i>Fed, BoJ, ECB...</i>"]
T18["bloomberg_get_earnings_calendar<br/><i>Earnings dates</i>"]
end
style T5 fill:#1a73e8,stroke:#1557b0,color:#fff
style T6 fill:#1a73e8,stroke:#1557b0,color:#fff
style T7 fill:#1a73e8,stroke:#1557b0,color:#fff
style T8 fill:#1a73e8,stroke:#1557b0,color:#fff
style T9 fill:#7b1fa2,stroke:#6a1b9a,color:#fff
style T10 fill:#7b1fa2,stroke:#6a1b9a,color:#fffWhat's New (vs upstream)
Modular architecture — server.py refactored from 1,798 lines to ~89 lines. Handlers, models, formatters, and utils cleanly separated.
9 new tools — BDS bulk data, multi-period estimates, technical analysis (//blp/tasvc), ownership analysis, supply chain (SPLC), BQL queries (//blp/bqlsvc), and more.
Cache layer — TTL-based in-memory cache with data-type-aware expiration (30s for prices, 24h for static data).
10 new FieldSets — Pre-defined field collections for estimates, profitability, cash flow, balance sheet, ownership, governance, risk, valuation, earnings surprise, and growth.
Full Bloomberg API surface — Covers //blp/refdata, //blp/instruments, //blp/apiflds, //blp/tasvc, //blp/bqlsvc.
Data Flow
sequenceDiagram
participant Client as AI Assistant
participant MCP as MCP Server
participant Cache as Cache Layer
participant Val as Pydantic Validation
participant Expand as Field Expander
participant Session as BloombergSession
participant BBG as Bloomberg Terminal
Client->>MCP: Tool call (JSON)
MCP->>Val: Validate input model
Val-->>MCP: Validated params
alt FieldSet shortcuts used
MCP->>Expand: Expand FieldSet shortcuts
Expand-->>MCP: Resolved field list
end
MCP->>Cache: Check cache
alt Cache hit
Cache-->>MCP: Cached result
else Cache miss
MCP->>Session: Build & send request
Session->>BBG: blpapi Request
BBG-->>Session: blpapi Response
Session-->>MCP: Parsed dataclasses
MCP->>Cache: Store with TTL
end
alt Markdown format
MCP-->>Client: Formatted table
else JSON format
MCP-->>Client: Structured JSON
endBloomberg Services
graph LR
subgraph "Bloomberg Terminal (localhost:8194)"
R["//blp/refdata<br/><i>BDP, BDH, BDS, BEQS</i>"]
I["//blp/instruments<br/><i>Security lookup</i>"]
F["//blp/apiflds<br/><i>Field discovery</i>"]
T["//blp/tasvc<br/><i>Technical analysis</i>"]
B["//blp/bqlsvc<br/><i>Query language</i>"]
end
R --- T1[Reference Data]
R --- T2[Historical Data]
R --- T3[Intraday Bars/Ticks]
R --- T4[Bulk Data]
R --- T5[Estimates]
R --- T6[Screening]
I --- T7[Security Search]
F --- T8[Field Search]
T --- T9[RSI / MACD / Bollinger]
B --- T10[Dynamic Queries]
style R fill:#ff6f00,stroke:#e65100,color:#fff
style T fill:#7b1fa2,stroke:#6a1b9a,color:#fff
style B fill:#1a73e8,stroke:#1557b0,color:#fffTool Reference
Market Data
Tool | Description | Key Parameters |
| Current field values (BDP) for any security |
|
| Time series (BDH) with configurable periodicity |
|
| OHLCV candles (1/5/15/30/60 min) |
|
| Raw tick-level trade/quote data |
|
Bulk Data & Analysis — NEW
Tool | Description | Key Parameters |
| Bulk reference data (BDS) — holders, dividends, supply chain, index members |
|
| Multi-period consensus estimates with revision momentum |
|
| Comprehensive ownership analysis (holders + insider + institutional) |
|
| Bloomberg SPLC supply chain data (suppliers, customers, competitors) |
|
Technical Analysis & BQL — NEW
Tool | Description | Key Parameters |
| TA indicators via //blp/tasvc (RSI, MACD, Bollinger, SMA, EMA, DMI, Stochastic) |
|
| Execute Bloomberg Query Language queries |
|
Discovery
Tool | Description | Key Parameters |
| Find securities by name or partial ticker |
|
| Discover Bloomberg field mnemonics |
|
| Detailed field metadata and documentation |
|
Screening
Tool | Description | Key Parameters |
| Execute saved Bloomberg EQS screens |
|
| Index/screen constituents with optional fields |
|
| Custom filtering, ranking, and field selection |
|
Calendars
Tool | Description | Key Parameters |
| Upcoming macro releases by region/importance |
|
| Earnings announcements by universe/timing |
|
All tools support response_format: "markdown" (default) or "json".
FieldSet Shortcuts
Instead of remembering Bloomberg field mnemonics, use shorthand names that expand to multiple fields.
Core FieldSets
FieldSet | Fields | Description |
| 5 | PX_LAST, PX_OPEN, PX_HIGH, PX_LOW, CHG_PCT_1D |
| 4 | CHG_PCT_1D, CHG_PCT_5D, CHG_PCT_1M, CHG_PCT_YTD |
| 7 | + CHG_PCT_3M, CHG_PCT_6M, CHG_PCT_1YR |
| 3+1 | VOLUME, VOLUME_AVG_20D, TURNOVER + computed rvol |
| 4 | RSI_14D, VOLATILITY_30D, VOLATILITY_90D, BETA_RAW_OVERRIDABLE |
| 5 | PE_RATIO, PX_TO_BOOK_RATIO, EV_TO_EBITDA, DIVIDEND_YIELD, CUR_MKT_CAP |
| 3 | EQY_REC_CONS, BEST_TARGET_PRICE, BEST_EPS |
| 2 | GICS_SECTOR_NAME, GICS_INDUSTRY_NAME |
| 30+ | All of the above combined |
Analytical FieldSets — NEW
FieldSet | Fields | Key Bloomberg Mnemonics |
| 10 | BEST_EPS, BEST_SALES, BEST_EPS_4WK_CHG, BEST_TARGET_PRICE |
| 7 | GROSS_MARGIN, ROE, ROA, ROIC, OPER_MARGIN |
| 6 | FCF_YIELD, CF_FROM_OPS, NET_INCOME, EBITDA |
| 6 | D/E, INTEREST_COV, CUR_RATIO, NET_DEBT |
| 5 | INSIDER%, INST%, SHORT_INT_RATIO |
| 4 | ESG scores (overall, E, S, G) |
| 6 | BETA, VOL 10/30/90/260D, MKT_CAP |
| 9 | PE, P/B, P/S, EV/EBITDA, P/FCF, DVD_YLD |
| 6 | EPS/sales actual vs estimate + surprise |
| 4 | Sales/EPS/EBITDA growth, LT growth est |
Dynamic Screening
The most powerful tool. Build custom screens with pre-validated field sets, filters, and ranking — no need to know Bloomberg field mnemonics.
flowchart LR
A["Universe<br/><i>index, screen,<br/>or ticker list</i>"] --> B["Field Expansion<br/><i>FieldSet shortcuts<br/>→ Bloomberg fields</i>"]
B --> C["Bloomberg API<br/><i>ReferenceDataRequest</i>"]
C --> D["Filter<br/><i>gt, lt, between,<br/>in, eq, ...</i>"]
D --> E["Rank & Slice<br/><i>rank_by + top_n</i>"]
E --> F["Response<br/><i>Markdown table<br/>or JSON</i>"]
style A fill:#e8f5e9,stroke:#2e7d32
style C fill:#fff3e0,stroke:#ff6f00
style F fill:#e3f2fd,stroke:#1a73e8Filter Operators
Operator | Description | Example |
| Greater than (or equal) |
|
| Less than (or equal) |
|
| Equals / not equals |
|
| Range (inclusive) |
|
| Value in list |
|
Example: Find Oversold High-Volume Stocks
{
"universe": "index:SPX Index",
"fields": ["PRICE", "RVOL", "TECHNICAL", "SECTOR"],
"filters": [
{"field": "RSI_14D", "op": "lt", "value": 30},
{"field": "rvol", "op": "gt", "value": 2.0}
],
"rank_by": "rvol",
"rank_descending": true,
"top_n": 20
}Common BDS (Bulk Data) Fields
Field | Returns |
| Top 20 shareholders with positions and dates |
| Complete dividend history |
| Supplier list with revenue exposure |
| Customer list with revenue exposure |
| Competitor list |
| Index constituents |
| Analyst ratings detail |
| Historical earnings with actual EPS |
| Board members |
Cache Layer
The built-in cache reduces Bloomberg API load with data-type-aware TTLs:
Data Type | TTL | Rationale |
Static reference (name, sector) | 24 hours | Rarely changes |
Financial statements | 7 days | Quarterly updates |
Estimates / consensus | 4 hours | Updates throughout day |
Price / volume | 30 seconds | Near real-time |
Historical (EOD) | 12 hours | End-of-day data stable |
Bulk data (holders, supply chain) | 24 hours | Daily updates |
Project Structure
graph TB
subgraph "src/bloomberg_mcp/"
SERVER["server.py<br/><i>~89 lines, thin entry point</i>"]
subgraph models["models/"]
INPUTS["inputs.py<br/><i>22 Pydantic models</i>"]
ENUMS["enums.py<br/><i>ResponseFormat, modes</i>"]
end
subgraph handlers["handlers/"]
direction TB
H_REF["reference.py"]
H_HIST["historical.py"]
H_INTRA["intraday.py"]
H_SEARCH["search.py"]
H_SCREEN["screening.py"]
H_CAL["calendars.py"]
H_BULK["bulk.py ★"]
H_EST["estimates.py ★"]
H_TECH["technical.py ★"]
H_OWN["ownership.py ★"]
H_SC["supply_chain.py ★"]
H_BQL["bql.py ★"]
end
FMTR["formatters.py"]
UTILS["utils.py"]
subgraph core["core/"]
SESSION["session.py<br/><i>Singleton</i>"]
CACHE["cache.py ★<br/><i>TTL cache</i>"]
REQ["requests.py"]
RESP["responses.py"]
end
subgraph tools["tools/"]
direction TB
T_REF["reference.py"]
T_HIST["historical.py"]
T_SEARCH["search.py"]
subgraph ds["dynamic_screening/"]
MODELS_DS["models.py<br/><i>19+ FieldSets</i>"]
SCREEN_DS["screen.py"]
FILTERS["filters.py"]
end
subgraph mn["morning_note/"]
MN_CFG["config.py"]
MN_US["us_session.py"]
MN_STORE["storage.py"]
end
end
end
SERVER --> handlers
SERVER --> models
handlers --> core
handlers --> FMTR
handlers --> UTILS
style SERVER fill:#1a73e8,stroke:#1557b0,color:#fff
style SESSION fill:#2e7d32,stroke:#1b5e20,color:#fff
style CACHE fill:#2e7d32,stroke:#1b5e20,color:#fff
style H_BULK fill:#e65100,stroke:#bf360c,color:#fff
style H_EST fill:#e65100,stroke:#bf360c,color:#fff
style H_TECH fill:#e65100,stroke:#bf360c,color:#fff
style H_OWN fill:#e65100,stroke:#bf360c,color:#fff
style H_SC fill:#e65100,stroke:#bf360c,color:#fff
style H_BQL fill:#e65100,stroke:#bf360c,color:#fff★ = New in v1.1
Installation
Prerequisites
Python 3.10+
Bloomberg Terminal running and logged in — connects via localhost:8194
Setup
# 1. Install Bloomberg Python SDK
pip install blpapi
# 2. Install bloomberg-mcp
git clone https://github.com/QmQsun/Bloomberg-MCP.git
cd Bloomberg-MCP
pip install . # standard install
# or: pip install -e . # editable mode (for development)Note:
blpapi3.19.0+ ships pre-built wheels —pip install blpapiworks directly on Windows, macOS, and Linux without additional setup.If
pip install blpapifails (older platforms or Python versions), install via the C++ SDK:# Set Bloomberg C++ SDK path export BLPAPI_ROOT=/path/to/blpapi_cpp_3.x.x.x # Linux/macOS set BLPAPI_ROOT=C:\blp\blpapi_cpp_3.x.x.x # Windows pip install blpapi
Configure Claude Code
Add to your Claude Code MCP settings:
{
"mcpServers": {
"bloomberg-mcp": {
"command": "python",
"args": ["-m", "bloomberg_mcp.server"],
"cwd": "/path/to/bloomberg-mcp",
"env": {
"BLOOMBERG_HOST": "localhost",
"BLOOMBERG_PORT": "8194"
}
}
}
}Configure GPT Codex
Add to your codex config.toml
[mcp_servers.bloomberg-mcp]
command = "python"
args = ["-m", "bloomberg_mcp.server"]
cwd = "/path/to/bloomberg-mcp"
[mcp_servers.bloomberg-mcp.env]
BLOOMBERG_HOST = "localhost"
BLOOMBERG_PORT = "8194"if it cannot load to codex, check your cwd path
Quick Start
As a Python Library
from bloomberg_mcp.tools import get_reference_data, get_historical_data
# Current prices and fundamentals
data = get_reference_data(
securities=["AAPL US Equity", "700 HK Equity", "2899 HK Equity"],
fields=["PX_LAST", "PE_RATIO", "DIVIDEND_YIELD"]
)
for sec in data:
print(f"{sec.security}: {sec.fields.get('PX_LAST')}")
# Multi-period estimates with overrides
data = get_reference_data(
securities=["CEG US Equity"],
fields=["BEST_EPS", "BEST_EPS_4WK_CHG", "BEST_TARGET_PRICE"],
overrides={"BEST_FPERIOD_OVERRIDE": "1FY"}
)
# Historical time series
hist = get_historical_data(
securities=["SPY US Equity"],
fields=["PX_LAST", "VOLUME"],
start_date="20240101",
end_date="20241231",
periodicity="DAILY"
)As an MCP Server
# stdio (default — for Claude Code)
python -m bloomberg_mcp.server
# HTTP transport (for web clients)
python -m bloomberg_mcp.server --http --port=8080
# SSE transport (for streaming clients)
python -m bloomberg_mcp.server --sse --port=8080Docker Deployment
graph LR
subgraph Docker
MCP["bloomberg-mcp<br/><i>:8080</i>"]
end
subgraph Host
BBG["Bloomberg Terminal<br/><i>:8194</i>"]
end
Client["AI Client"] -- "HTTP/SSE" --> MCP
MCP -- "host.docker.internal:8194" --> BBG
style MCP fill:#1a73e8,stroke:#1557b0,color:#fff
style BBG fill:#ff6f00,stroke:#e65100,color:#fffdocker-compose up -d
docker-compose logs -f bloomberg-mcpSecurity Identifier Formats
AAPL US Equity # US stock
VOD LN Equity # UK stock (London)
7203 JP Equity # Japan stock (numeric)
700 HK Equity # Hong Kong stock
1133 HK Equity # Hong Kong stock (numeric)
601012 CH Equity # A-share (Shanghai)
002594 CH Equity # A-share (Shenzhen)
300750 CH Equity # A-share (ChiNext)
SPX Index # Index
HSI Index # Hang Seng Index
VIX Index # Volatility index
EUR Curncy # Currency
CL1 Comdty # Commodity future
SPY US Equity # ETFContributing
Contributions welcome! Please open an issue or submit a pull request.
pip install -e ".[dev]"
pytest # Unit tests (no Bloomberg needed)
pytest tests/integration/ # Integration tests (needs Terminal)
black src/ tests/
ruff check src/ tests/Contributors
Contributor | Role | |
1 | Architecture refactor, 6 new tools, caching layer, FieldSets, code review | |
2 | Bug discovery ( | |
3 | Claude (Anthropic) | Implementation assistance, code generation, QA |
4 | Original project foundation |
License
MIT — see LICENSE for details.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/sarat-ayon/Bloomberg-MCP'
If you have feedback or need assistance with the MCP directory API, please join our Discord server