docs

BaseAgent Documentation

Professional documentation for the BaseAgent autonomous coding assistant

BaseAgent is a high-performance autonomous agent designed for the Term Challenge. It leverages LLM-driven decision making with advanced context management and cost optimization techniques.

Challenge API policy: this agent is configured to use only the DeepSeek API for cost reasons. Challenge runs must use DEEPSEEK_API_KEY and the configured DeepSeek model. Do not add or rely on Chutes, OpenRouter, Anthropic, OpenAI, or other provider fallbacks for challenge execution.

---

Table of Contents

Getting Started

• Overview - What is BaseAgent and core design principles
• Installation - Prerequisites and setup instructions
• Quick Start - Your first task in 5 minutes

Core Concepts

• Architecture - Technical architecture and system design
• Configuration - All configuration options explained
• Usage Guide - Command-line interface and options

Reference

• Tools Reference - Available tools and their parameters
• Context Management - Token management and compaction
• Best Practices - Optimal usage patterns

LLM Provider

• DeepSeek API Integration - Using DeepSeek as the challenge API

---

Quick Navigation

Document	Description
Overview	High-level introduction and design principles
Installation	Step-by-step setup guide
Quick Start	Get running in minutes
Architecture	Technical deep-dive with diagrams
Configuration	Environment variables and settings
Usage	CLI commands and examples
Tools	Complete tools reference
Context Management	Memory and token optimization
Best Practices	Tips for optimal performance
DeepSeek Integration	DeepSeek API setup and usage

---

Architecture at a Glance

graph TB
    subgraph User["User Interface"]
        CLI["CLI (agent.py)"]
    end
    
    subgraph Core["Core Engine"]
        Loop["Agent Loop"]
        Context["Context Manager"]
        Cache["Prompt Cache"]
    end
    
    subgraph LLM["LLM Layer"]
        Client["DeepSeek HTTP Client"]
        Provider["DeepSeek API"]
    end
    
    subgraph Tools["Tool System"]
        Registry["Tool Registry"]
        Shell["shell_command"]
        Files["read_file / write_file"]
        Search["grep_files / list_dir"]
    end
    
    CLI --> Loop
    Loop --> Context
    Loop --> Cache
    Loop --> Client
    Client --> Provider
    Loop --> Registry
    Registry --> Shell
    Registry --> Files
    Registry --> Search

Loading

---

Key Features

• Fully Autonomous - No user confirmation required; makes decisions independently
• LLM-Driven - All decisions made by the language model, not hardcoded logic
• Prompt Caching - 90%+ cache hit rate for significant cost reduction
• Context Management - Intelligent pruning and compaction for long tasks
• Self-Verification - Automatic validation before task completion
• DeepSeek Only for Challenge Runs - Uses DEEPSEEK_API_KEY, https://api.deepseek.com, provider deepseek, and deepseek-v4-pro

---

Project Structure

baseagent/
├── agent.py                 # Entry point
├── src/
│   ├── core/
│   │   ├── loop.py          # Main agent loop
│   │   └── compaction.py    # Context management
│   ├── llm/
│   │   └── client.py        # LLM client (DeepSeek/httpx)
│   ├── config/
│   │   └── defaults.py      # Configuration
│   ├── tools/               # Tool implementations
│   ├── prompts/
│   │   └── system.py        # System prompt
│   └── output/
│       └── jsonl.py         # JSONL event emission
├── rules/                   # Development guidelines
├── astuces/                 # Implementation techniques
└── docs/                    # This documentation

---

License

MIT License - See LICENSE for details.