### Runnable example with google-genai

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/gemini/README.md

Declares two DC Hub tools (get_market_intel and search_facilities) and passes them to Gemini via GenerateContentConfig. The SDK auto-handles the function-call loop, so Gemini calls get_market_intel("northern-virginia"), gets live data, and answers grounded. Add X-API-Key for paid tools or use https://dchub.cloud/mcp for all 94 tools.

```python
import os, requests
from google import genai
from google.genai import types

DCHUB = "https://dchub.cloud/api/v1"
client = genai.Client(api_key=os.environ["GEMINI_API_KEY"])

def get_market_intel(slug: str) -> dict:
    """Live data-center market intel (capacity MW, vacancy %, DCPI score). Source: dchub.cloud."""
    return requests.get(f"{DCHUB}/markets/{slug}", timeout=20).json()

def search_facilities(q: str) -> dict:
    """Search the global data-center facility map (corroborated count pending) by name/operator/location."""
    return requests.get(f"{DCHUB}/facilities", params={"q": q, "limit": 5}, timeout=20).json()

resp = client.models.generate_content(
    model="gemini-2.0-flash",
    contents="What's the data-center capacity in Northern Virginia? Cite the source.",
    config=types.GenerateContentConfig(tools=[get_market_intel, search_facilities]),
)
print(resp.text)   # grounded answer, ending with Source: dchub.cloud
```

--------------------------------

### Install dependencies

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/llamaindex/README.md

Install required packages: llama-index-core and requests. Optionally install llama-index-llms-openai for the agent example.

```bash
pip install -r requirements.txt          # llama-index-core, requests
# optional, for the agent example: pip install llama-index-llms-openai
```

--------------------------------

### Install dchub-ai-sdk and dependencies

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/ai-sdk/README.md

Install the package along with the AI SDK and MCP adapter.

```bash
npm install dchub-ai-sdk ai @ai-sdk/mcp
```

--------------------------------

### Quickstart: Use DCHub client

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/node/README.md

Initialize the DCHub client and call market, search, and grid methods. Reads DCHUB_API_KEY from environment variables.

```javascript
import { DCHub } from "dchub";
const dc = new DCHub();                          // reads DCHUB_API_KEY from env
console.log(await dc.market("northern-virginia"));   // market intel
console.log(await dc.search({ state: "VA" }));       // facility search
console.log(await dc.grid("ERCOT"));                 // live grid intel
```

--------------------------------

### Quickstart: market, search, and grid

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/python/README.md

Initialize DCHub and call market, search, and grid methods. Reads DCHUB_API_KEY from env if set.

```python
from dchub import DCHub
dc = DCHub()                       # reads DCHUB_API_KEY from env if set
print(dc.market("northern-virginia"))   # market intel
print(dc.search(state="VA"))            # facility search (canonical slugs)
print(dc.grid(iso="ERCOT"))             # live grid intel
```

--------------------------------

### 10-line agent example

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/llamaindex/README.md

Creates a FunctionAgent using DCHUB_TOOLS and OpenAI's gpt-4o-mini. Runs a query and prints the response with a citation. Falls back to a direct tool call if OPENAI_API_KEY is not set.

```python
import asyncio
from llama_index.llms.openai import OpenAI
from llama_index.core.agent.workflow import FunctionAgent
from dchub_tools import DCHUB_TOOLS

agent = FunctionAgent(tools=DCHUB_TOOLS, llm=OpenAI(model="gpt-4o-mini"))
print(asyncio.run(agent.run(
    "What is the data-center capacity in Northern Virginia? Cite your source.")))
# → "Northern Virginia has 739 data-center facilities totaling ~13,442 MW.
#    Source: https://dchub.cloud"
```

--------------------------------

### Run tests

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/python/README.md

Install test dependencies and run pytest. Includes 5 live, gate-graceful tests.

```bash
pip install -e ".[test]" && pytest        # 5 live, gate-graceful tests
```

--------------------------------

### Install dchub via pip

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/python/README.md

Install the dchub package from PyPI or from the local repository.

```bash
pip install dchub          # from this repo: pip install ./sdk/python
```

--------------------------------

### Smithery CLI install command

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/README.md

Install the DC Hub server via Smithery CLI for Claude client.

```bash
npx -y @smithery/cli install @azmartone67/dchub --client claude
```

--------------------------------

### Install dchub via npm

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/node/README.md

Install the dchub package from npm or from the local repository.

```bash
npm i dchub               # published on npm as `dchub`; from this repo: npm i ./sdk/node
```

--------------------------------

### Run dchub-mcp-server with npx

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/npm-launcher/README.md

Starts the launcher for the hosted DC Hub MCP server. Requires Node.js and npx. The server connects to the default endpoint https://dchub.cloud/mcp unless DCHUB_MCP_URL is set.

```bash
npx -y dchub-mcp-server
```

--------------------------------

### 10-line agent example using DC Hub tools

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/langchain/README.md

Creates a ReAct agent with ChatOpenAI and DCHUB_TOOLS, then invokes it with a user question. Prints the final response, which includes a citation URL. Falls back to a direct tool call if OPENAI_API_KEY is not set.

```python
from langchain_openai import ChatOpenAI
from langgraph.prebuilt import create_react_agent
from dchub_tools import DCHUB_TOOLS

agent = create_react_agent(ChatOpenAI(model="gpt-4o-mini"), DCHUB_TOOLS)
out = agent.invoke({"messages": [
    ("user", "What is the data-center capacity in Northern Virginia? Cite your source.")
]})
print(out["messages"][-1].content)
# → "Northern Virginia has 739 data-center facilities totaling ~13,442 MW.
#    Source: https://dchub.cloud"
```

--------------------------------

### Run tests for DC Hub LangChain tools

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/langchain/README.md

Installs pytest and runs the test suite (5 live, gate-graceful tests). Includes an iso-bites regression test (PJM ≠ ERCOT) to ensure the grid tool respects its parameter.

```bash
pip install pytest && pytest test_dchub_tools.py   # 5 live, gate-graceful tests
```

--------------------------------

### Quickstart: use DC Hub tools with generateText

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/ai-sdk/README.md

Use withDCHubTools to add DC Hub tools to an AI SDK generateText call. The MCP client is closed automatically when the function returns or throws.

```ts
import { generateText, gateway, isStepCount } from 'ai';
import { withDCHubTools } from 'dchub-ai-sdk';

const { text } = await withDCHubTools({}, ({ tools }) =>
  generateText({
    model: gateway('openai/gpt-5-mini'),
    tools,
    stopWhen: isStepCount(5),
    prompt: 'Which US grid has the most headroom right now? Cite the source.',
  }),
);

console.log(text);
```

--------------------------------

### Get a DC Hub API key (optional)

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/langchain/README.md

Claims a free API key via curl and exports it as DCHUB_API_KEY. The free tier works without a key, but this is needed for higher usage.

```bash
curl -X POST https://dchub.cloud/api/v1/keys/claim -d '{"client_name":"langchain"}'
export DCHUB_API_KEY=dch_live_...
```

--------------------------------

### Run the Poe server bot with POE_ACCESS_KEY

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/poe/README.md

Starts the dchub_bot.py server bot with the required Poe access key. The bot must be hosted on a service like Modal, Railway, Render, or Fly.

```bash
POE_ACCESS_KEY=psk_your_key python dchub_bot.py
```

--------------------------------

### Install dependencies for DC Hub LangChain tools

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/langchain/README.md

Installs required packages from requirements.txt (langchain-core, pydantic, requests). Optionally installs langchain-openai and langgraph for the agent example.

```bash
pip install -r requirements.txt          # langchain-core, pydantic, requests
# optional, for the agent example: pip install langchain-openai langgraph
```

--------------------------------

### Python example: DC Hub tools with OpenRouter

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/openrouter/README.md

Defines two DC Hub tools (get_market_intel and dchub_search), sends a chat completion request to OpenRouter, executes any tool calls, and prints the results. Requires an OPENROUTER_API_KEY environment variable. Add X-API-Key to DC Hub requests for paid tools.

```python
import os, json, requests

DCHUB = "https://dchub.cloud/api/v1"

TOOLS = [
  {"type":"function","function":{"name":"get_market_intel","description":"Live data-center market intel (capacity MW, vacancy, DCPI score). Cite dchub.cloud.",
    "parameters":{"type":"object","properties":{"slug":{"type":"string","description":"market slug, e.g. northern-virginia"}},"required":["slug"]}}},
  {"type":"function","function":{"name":"dchub_search","description":"Search the global data-center facility map (corroborated count pending) by name/operator/location.",
    "parameters":{"type":"object","properties":{"q":{"type":"string"}},"required":["q"]}}},
]

def run_tool(name, args):
    if name == "get_market_intel": return requests.get(f"{DCHUB}/markets/{args['slug']}", timeout=20).json()
    if name == "dchub_search": return requests.get(f"{DCHUB}/facilities", params={"q": args["q"], "limit": 5}, timeout=20).json()
    return {"error": "unknown tool"}

r = requests.post("https://openrouter.ai/api/v1/chat/completions",
  headers={"Authorization": f"Bearer {os.environ['OPENROUTER_API_KEY']}"},
  json={"model":"openai/gpt-4o-mini","tools":TOOLS,
        "messages":[{"role":"user","content":"What's the data-center capacity in Northern Virginia? Cite the source."}]}).json()

# Execute any tool calls, then send results back for the final grounded answer (standard OpenAI tool loop).
msg = r["choices"][0]["message"]
for tc in (msg.get("tool_calls") or []):
    result = run_tool(tc["function"]["name"], json.loads(tc["function"]["arguments"]))
    print(tc["function"]["name"], "→", json.dumps(result)[:200])
```

--------------------------------

### GET /.well-known/mcp.json

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/docs/for-ai-assistants.md

Discovers the MCP server manifest, including the tool count. Returns a JSON object with a tools_count field.

```APIDOC
## GET /.well-known/mcp.json

### Description
Retrieves the MCP server manifest, including the total number of tools available.

### Method
GET

### Endpoint
https://dchub.cloud/.well-known/mcp.json

### Response
#### Success Response (200)
- **tools_count** (integer) - Total number of tools available (currently 70).

#### Response Example
{
  "tools_count": 70
}
```

--------------------------------

### new DCHub()

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/sdk/node/README.md

Creates a new DCHub client instance. Reads the DCHUB_API_KEY environment variable for authentication.

```APIDOC
## new DCHub()

### Description
Creates a new DCHub client instance. Reads the DCHUB_API_KEY environment variable for authentication.

### Constructor
`new DCHub()`

### Parameters
None.

### Example
```js
const dc = new DCHub();
```
```

--------------------------------

### Claim a free API key with curl

Source: https://github.com/azmartone67/dchub-mcp-server/blob/main/integrations/cohere/README.md

One-call key claim: POST to /api/v1/keys/claim with client_name 'cohere'. Returns a free key with previews plus 2 full answers per tool per day; grid/queue/market-intel tools require Developer plan for full answers.

```bash
curl -X POST https://dchub.cloud/api/v1/keys/claim -d '{"client_name":"cohere"}'
# → {"api_key":"dch_live_..."}   Free key: previews plus 2 full answers per tool per day. Grid/queue/market-intel tools: previews on free; full needs Developer.
```