Warbot

CLI bot using OpenAI's gpt-5-mini with streaming, thinking display, tool calls, and multi-turn conversations focused on world conflicts, location-based risk assessment, and preparedness guidance.

Features

• Streaming responses with thinking tokens surfaced in the terminal
• Tool calls with a pluggable registry
• Conversation history across turns

Tools

Warbot includes three built-in tools for conflict awareness and preparedness:

get_world_conflicts

Fetches current world conflicts from Wikipedia, including:

• Major wars (10,000+ combat-related deaths)
• Minor wars (1,000-9,999 deaths)
• Conflicts (100-999 deaths)
• Skirmishes (<100 deaths)

Data is cached locally for 1 hour to reduce API load. Supports optional region filtering (e.g., "Europe", "Middle East", "Asia Pacific"). Returns conflict details including start date, location, continent, and fatality statistics.

assess_location_risks

Assesses risks for a specified location (city, country, or coordinates), evaluating:

• Proximity to active conflicts
• Infrastructure stability
• Digital/internet reliability
• Other relevant risk factors

Currently returns structured risk assessments with categories and severity levels. Note: This tool uses stubbed data; integrate geocoding and threat intelligence sources for production use.

get_preparation_guidance

Provides structured preparation guidance for emergency scenarios such as:

• Utilities interruption
• Internet loss
• Armed conflict
• Natural disasters

Returns actionable guidance organized by:

• Immediate actions: What to do right away
• Short-term: 72-hour preparedness steps
• Long-term: Extended preparation strategies
• Supplies: Essential items to have on hand
• Communication: How to stay connected
• Evacuation: Routes and go-bag preparation

Guidance is tailored based on the scenario type and optional location context.

Installation

python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -e .
# or, install via requirements
pip install -r requirements.txt

Configuration

Create a .env file (see .env.example):

OPENAI_API_KEY=your_api_key_here
# Optional
# OPENAI_MODEL=gpt-5-mini
# OPENAI_BASE_URL=

Usage

# After installation
warbot

# Using python module
python -m warbot

# Override model
warbot --model gpt-5-mini

# Enable debug logging of raw stream chunks
warbot --debug

# Provide a first question automatically
warbot --question "What major conflicts are active right now?"

Type exit or quit to leave the session.

Development

pip install -r requirements-dev.txt
pytest

Project Layout

src/warbot/
  __main__.py          # CLI entry
  bot.py               # Core bot logic
  stream_handler.py    # Streaming parser
  config.py            # Settings and client builder
  tools/               # Tool framework + tools
tests/                 # Test suite

Notes

• get_world_conflicts fetches real data from Wikipedia and caches it locally.
• assess_location_risks currently uses stubbed data; integrate geocoding and threat intelligence sources for production.
• Thinking and content are streamed; intermediate thoughts are shown in dim text.