CreativeAct Enterprise LangChain Prompts

This repository contains a collection of structured prompt templates designed for LangChain integration. The prompts cover enterprise use cases including Fraud Detection, Strategy, Operations, HR, and AI Governance.

Table of Contents

• CreativeAct Enterprise LangChain Prompts
  • Table of Contents
  • Structure
  • Setup
    • Venv Setup and Dependencies
  • Usage
    • Loading a Prompt
    • Using with an LLM (OpenAI)
    • Using with a Local LLM (Ollama)
  • 📚 Available Categories
  • 🤝 Contributing
• 📄 License

Structure

• prompts/: Contains JSON files categorized by domain. Each file holds an array of prompt objects.
• src/loader.py: A custom utility to load prompts into LangChain PromptTemplate objects.

Setup

Venv Setup and Dependencies

# create venv
python3 -m venv venv

# activate
.venv/scripts/activate

pip install -r requirements.txt

Usage

Loading a Prompt

Load a specific prompt by its ID and format it with your data.

from src.loader import PromptRepoLoader

# 1. Initialize the loader pointing to the 'prompts' folder
loader = PromptRepoLoader("./prompts")

# 2. Load a specific prompt (e.g., FRAUD-001 from the Fraud category)
prompt = loader.get_prompt_by_id("fraud_detection_security.json", "FRAUD-001")

# 3. Define your input variables
inputs = {
    "new_claim": "My vehicle was stolen from the parking lot."
}

# 4. Format the prompt
formatted_text = prompt.format(**inputs)

print(formatted_text)

Using with an LLM (OpenAI)

If you have an OpenAI API key, you can connect these prompts directly to a model.

import os
from langchain_openai import ChatOpenAI
from src.loader import PromptRepoLoader

# Set your API Key
os.environ["OPENAI_API_KEY"] = "sk-your-key-here"

loader = PromptRepoLoader("./prompts")
prompt = loader.get_prompt_by_id("knowledge_management.json", "KM-002")

# Initialize the model
model = ChatOpenAI(model="gpt-3.5-turbo")

# Create a chain: Prompt -> Model
chain = prompt | model

# Invoke the chain
response = chain.invoke({
    "PROJECT_NAME": "Project Phoenix",
    "PROJECT_ARTIFACTS": "We missed the deadline because the API key expired."
})

print(response.content)

Using with a Local LLM (Ollama)

For air-gapped environments or local execution, you can use these prompts with Ollama.

Prerequisites:

1. Install and run Ollama.
2. Pull your desired model (e.g., ollama pull llama3).
3. Ensure the Ollama service is running (usually on http://localhost:11434).
4. Install the LangChain Ollama integration: pip install langchain-ollama.

Implementation:

from langchain_ollama import ChatOllama
from src.loader import PromptRepoLoader

# 1. Initialize the loader
loader = PromptRepoLoader("./prompts")
prompt = loader.get_prompt_by_id("strategy_market_intelligence.json", "STRAT-001")

# 2. Initialize the Local Ollama Model
# Ensure you have pulled the model (e.g., 'llama3' or 'mistral') via Ollama CLI
model = ChatOllama(model="llama3", base_url="http://localhost:11434")

# 3. Create a chain
chain = prompt | model

# 4. Invoke the chain
response = chain.invoke({
    "COMPETITOR_DATA": "Competitor X has lowered prices by 20%.",
    "MARKET_CONTEXT": "Inflation is rising, consumer spending is down."
})

print(response.content)

📚 Available Categories

Category File	Content Description	Prompt IDs
fraud_detection_security.json	Coordinated fraud, insider threats, scam detection.	FRAUD-001 to FRAUD-015
risk_compliance_governance.json	Project risk, cybersecurity posture, ethical red teaming.	RISK-001 to RISK-005
strategy_market_intelligence.json	Competitive analysis, market entry, ESG initiatives.	STRAT-001 to STRAT-004
product_innovation.json	Feature prioritization, innovation bottlenecks.	PROD-001 to PROD-002
marketing_sales.json	Customer personas, objection handling.	MKT-001 to MKT-002
operations_process.json	Process optimization, cost management, customer journeys.	OPS-001 to OPS-005
hr_talent.json	Engagement surveys, skills gaps, burnout analysis.	HR-001 to HR-003
technology_ai.json	Tech adoption, AI readiness, human-AI workflows.	TECH-001 to TECH-003
knowledge_management.json	Knowledge ecosystems, lessons learned, taxonomy.	KM-001 to KM-005

🤝 Contributing

To add a new prompt:

1. Open the relevant JSON file in the prompts/ directory.
2. Add a new object following the existing schema (ensure you assign a unique id).

   {
  "id": "UNIQUE-ID",
  "name": "Prompt Name",
  "description": "Brief description of the use case.",
  "tags": ["relevant", "tags"],
  "input_variables": ["var1", "var2"],
  "template": "Your prompt text with {var1} placeholders.",
  "template_format": "f-string"
}

1. Validate your JSON syntax before committing.
2. Submit a pull request.

---

📄 License

This project is licensed under Apache 2.0 - see the LICENSE file for details.

@ 2026 CreativeAct Technologies