xrtm-forecast 0.2.1

Institutional-grade modular engine for generative forecasting and agentic reasoning.

xrtm-forecast

Institutional-grade parallelized agentic reasoning engine.

Overview

xrtm-forecast is an institutional-grade, domain-agnostic intelligence engine. It provides a framework for:

• Inference Layer: Standardized provider interfaces for Gemini, OpenAI, and local Hugging Face models.
• Tiered Reasoning: Composite RoutingAgent for cost-optimized task dispatching.
• Reasoning Graph: A pluggable state-machine orchestrator for multi-agent workflows.
• Agent Core: Standardized Agent base class for structured reasoning and parsing.
• Skill Protocol: Composable behaviors (e.g., Search, SQL, Pandas) that agents can dynamically equip.
• Observability: OTel-native structured telemetry and institutional execution reports.
• Evaluation: Built-in Backtest Engine for large-scale accuracy metrics (Brier Score, ECE).
• Temporal Sandboxing: Multi-layered protection (PiT Tools, Clock Mocking, Leakage Guardian) to prevent look-ahead bias in historical backtests.

Architectural Design: "Pure Core / The Kit"

xrtm-forecast is designed for modularity using a strict three-tier architecture:

• The Core (/core): Zero-dependency protocol layer and state-machine orchestrator.
• The Kit (/kit): Importable "Instruments" (Agents, Skills, Evaluators) for standard use cases.
• Providers (/providers): Concrete connectors for cloud and local inference backends.

Installation

From PyPI (Stable)

# Standard Institutional Install (Cloud + Core)
pip install xrtm-forecast

# Hardware-Specific Local Inference
pip install "xrtm-forecast[transformers]"  # PyTorch + HuggingFace
pip install "xrtm-forecast[vllm]"          # High-throughput serving
pip install "xrtm-forecast[llama-cpp]"     # CPU-optimized GGUF
pip install "xrtm-forecast[xlm]"           # Local Encoder specialists

# Researcher Kit (Enhanced Data/Viz)
pip install "xrtm-forecast[data,viz,memory]"

From Source (Latest)

pip install git+https://github.com/xrtm-org/forecast.git

Configuration

xrtm-forecast follows a decentralized configuration pattern. Global environment variables are used for infrastructure (API keys), while specific behaviors are controlled via module-level configuration classes.

1. Environment Secrets

Set your API keys in a .env file or environment:

# Core API Keys
GEMINI_API_KEY=your_key_here
OPENAI_API_KEY=your_key_here
TAVILY_API_KEY=your_key_here

2. Component Configuration

Each major module (inference, graph, telemetry, tools) has its own config.py defining its schema. This allows you to instantiate multiple components with different settings in the same process.

Quick Start

xrtm-forecast is designed for high-end ergonomics. Use the pre-configured assistants to start forecasting in seconds:

import asyncio
from forecast import create_forecasting_analyst

async def main():
    # 1. Instantiate the analyst with a shortcut
    # (API keys are automatically injected from your .env file)
    agent = create_forecasting_analyst(model_id="gemini")
    
    # 2. Execute reasoning on a complex probabilistic question
    result = await agent.run(
        "Will a general-purpose AI (AGI) be publicly announced before 2030?"
    )
    
    print(f"Confidence: {result.confidence}")
    print(f"Reasoning: {result.reasoning}")

if __name__ == "__main__":
    asyncio.run(main())

Documentation & Examples

• Architecture: The "Lego" Design
• Agent Registry: Pre-built & Core Agents
• Examples: Check the examples/ directory:
  • examples/kit/minimal_agent.py: One-line agent setup.
  • examples/kit/local_analyst.py: Private reasoning with HF models.
  • examples/kit/features/tiered_reasoning.py: Optimal routing between Fast/Smart tiers.
  • examples/kit/features/enterprise_data.py: Integrated SQL and Pandas analytics.
  • examples/kit/features/discovery.py: Dynamic skill discovery.
  • examples/kit/backtest_workflow.py: Historical backtesting with temporal sandboxing.

Contributing

We welcome institutional-grade contributions! Please see CONTRIBUTING.md for setup instructions and our development workflow.

---

License

xrtm-forecast is open-source software licensed under the Apache-2.0 license. See the LICENSE file for more details.

Copyright © 2026 XRTM Team.