Skip to main content

Installation Guide

Detailed installation instructions for PrecogX SDK and integrations.

System Requirements​

Python​

  • Python 3.8 or higher
  • pip package manager

Node.js​

  • Node.js 16.0 or higher
  • npm or yarn package manager

Other Languages​

  • REST API support for any language with HTTP capabilities

Python Installation​

Standard Installation​

pip install precogx-sdk

With Framework Support​

# For LangChain integration
pip install "precogx-sdk[langchain]"

# For AutoGen integration  
pip install "precogx-sdk[autogen]"

# For CrewAI integration
pip install "precogx-sdk[crewai]"
# Create virtual environment
python -m venv precogx-env

# Activate virtual environment
# On Windows:
precogx-env\Scripts\activate
# On macOS/Linux:
source precogx-env/bin/activate

# Install PrecogX SDK
pip install precogx-sdk

Verify Installation​

import precogx_sdk
print(f"PrecogX SDK version: {precogx_sdk.__version__}")

Node.js Installation​

Using npm​

npm install @precogx/sdk

Using yarn​

yarn add @precogx/sdk

Verify Installation​

const { PrecogXClient } = require('@precogx/sdk');
console.log('PrecogX SDK installed successfully');

Framework Integrations​

LangChain​

from precogx_sdk import PrecogXEmitter
from precogx_sdk.langchain import PrecogXCallbackHandler
from langchain.llms import OpenAI

# Set agent_id once on the emitter — the callback handler inherits it
emitter = PrecogXEmitter(
    backend_url="https://api.precogx.ai",
    api_key="your_api_key",
    agent_id="my-langchain-agent",    # set once here
)
handler = PrecogXCallbackHandler(emitter=emitter, agent_id=emitter.agent_id)

llm = OpenAI(temperature=0)
llm.call("Hello world", callbacks=[handler])

AutoGen​

from precogx_sdk.autogen import PrecogXAgent

agent = PrecogXAgent(
    name="assistant",
    api_key="your_api_key",
    agent_id="my-autogen-agent",      # identifies this agent in PrecogX dashboard
    system_message="You are a helpful assistant."
)

CrewAI​

from precogx_sdk.crewai import PrecogXAgent

agent = PrecogXAgent(
    role="Researcher",
    goal="Research topics",
    backstory="You are a research assistant",
    api_key="your_api_key",
    agent_id="my-crewai-researcher",  # identifies this agent in PrecogX dashboard
)

No-Code Platforms​

For no-code platforms, include agentId as a fixed value in your HTTP request body. Use a descriptive name — it will appear as the agent name in your PrecogX dashboard.

Flowise​

  1. Add HTTP Request node
  2. Set URL: https://api.precogx.ai/api/v1/telemetry/ingest
  3. Set method to POST
  4. Add header: x-api-key: YOUR_API_KEY
  5. Set JSON body:
{
  "agentId": "my-flowise-agent",
  "prompt": "{{input}}",
  "response": "{{output}}"
}

n8n​

  1. Add HTTP Request node
  2. Set URL: https://api.precogx.ai/api/v1/telemetry/ingest
  3. Set method to POST
  4. Add header: x-api-key: YOUR_API_KEY
  5. Set JSON body:
{
  "agentId": "my-n8n-agent",
  "prompt": "{{ $json.prompt }}",
  "response": "{{ $json.response }}"
}

Dify​

  1. Navigate to Tools → Custom Tool
  2. Set endpoint: https://api.precogx.ai/api/v1/telemetry/ingest
  3. Set method to POST
  4. Add header: x-api-key: YOUR_API_KEY
  5. Include agentId as a fixed string in the request body

Environment Configuration​

The simplest integration — set two env vars and you're done. No agent_id in any code.

# .env
PRECOGX_API_KEY=your_api_key_here
PRECOGX_AGENT_ID=my-agent-name        # the SDK reads this automatically

With these set, the entire integration is:

from precogx_sdk import PrecogXClient

client = PrecogXClient()   # reads both vars from environment automatically
client.send_telemetry({"prompt": "...", "response": "..."})

Multiple Agents​

For multiple agents, create one client per agent. All agents share the same API key (same organisation), but each has its own agent_id:

from precogx_sdk import PrecogXClient

# All use the same API key, each has its own identity
support = PrecogXClient(api_key="your_api_key", agent_id="support-bot")
sales   = PrecogXClient(api_key="your_api_key", agent_id="sales-bot")
hr      = PrecogXClient(api_key="your_api_key", agent_id="hr-screener")

# Now each client tracks its own agent independently
support.send_telemetry({"prompt": "...", "response": "..."})
sales.send_telemetry({"prompt": "...", "response": "..."})

Node.js Configuration​

const { PrecogXClient } = require('@precogx/sdk');

const client = new PrecogXClient({
  apiKey: process.env.PRECOGX_API_KEY,   // from environment
  agentId: process.env.PRECOGX_AGENT_ID, // from environment
});

Docker Installation​

Python Dockerfile​

FROM python:3.11-slim

WORKDIR /app
COPY requirements.txt .
RUN pip install -r requirements.txt

COPY . .
CMD ["python", "app.py"]

requirements.txt​

precogx-sdk==1.0.0
fastapi==0.104.1
uvicorn==0.24.0

Troubleshooting​

Common Issues​

Import Error

# Ensure you're using the correct Python environment
which python
pip list | grep precogx

Authentication Error

# Verify your API key
client = PrecogXClient(api_key="your_key")
print(client.health_check())

Network Issues

# Check connectivity
import requests
response = requests.get("https://api.precogx.ai/health")
print(response.status_code)

Getting Help​

Next Steps​