Project Overview

This GitHub Acceleration Initiative project (path: gh_mirrors/v0s/v0-system-prompts-models-and-tools) provides prompt configurations for multiple AI models and tools used in autonomous vehicle systems development.

Prompt Engnieering Framework for Autonomous Systems

Core Components

• AI Agents: Defined in VSCode Agent/Prompt.txt as "specialized automation agents" witth cross-language and framework expertise
• Tool Integration: Implemented via Traycer AI/phase_mode_tools.json with file path resolution capabilities
• Model Templates: Includes architecture-specific prompts like VSCode Agent/gpt-5.txt

Design Principles

• Concise task definitions (3-4 points, ≤60 words) as specified in Traycer AI/phase_mode_tools.json
• Contextual element references using backticks for code repository components
• File path inclusion for precise resource location by AI agents

Autonomous Driving Application Scenarios

Perception System Prompt Example

As an autonomous perception agent, analyze these three files for lane detection implementation:
- [VSCode Agent/gpt-5-mini.txt](https://link.gitcode.com/i/2aeb75e5f61ff7e53d02b514385465de)
- [Traycer AI/phase_mode_prompts.txt](https://link.gitcode.com/i/aa64a0c12b88240f8c9f116ac976f74c)
- [Claude Code/claude-code-tools.json](https://link.gitcode.com/i/f2682c1d326616f2500bec17d8b45dda)
Output must include error handling and edge case resolution modules.

Decision System Prompt Template

Implement a lane-change decision agent based on [Comet Assistant/System Prompt.txt](https://link.gitcode.com/i/b540cc5244c7c94805eeb92db33da6da). Requirements:
1. Use quality validation workflow from [VSCode Agent/gpt-4o.txt](https://link.gitcode.com/i/4f39fbcaa44bf32ed4c59f8bbea4be33)
2. Follow agent specification format in [Traycer AI/plan_mode_tools.json](https://link.gitcode.com/i/b9707cd15402a1b70dc05fdcc96d5808)
3. Generate decision tree pseudocode with safety distance calculations

Implementation Process

Development Workflow

1. Select appropriate AI model based on use case (e.g., VSCode Agent/gemini-2.5-pro.txt or VSCode Agent/claude-sonnet-4.txt)
2. Configure sensitive parameters via Leap.new/tools.json
3. Write phase-specific prompts following Traycer AI/phase_mode_tools.json schema
4. Validate implementations using VSCode Agent/gpt-5.txt quality gate procedures

Tool Invocation Example

{
  "ai_agent": {
    "identifier": "Agent ADS-DECSN",
    "objective": "Autonomous decision system implementation",
    "instruction": "Implement safe distance maintenance algorithm using sensor data, adhering to coding standards in [VSCode Agent/Prompt.txt](https://link.gitcode.com/i/37239925ca816a38853c5b1e1577869d)"
  }
}

Best Practices

• Maintain brevity per Traycer AI/phase_mode_tools.json word limits
• Specify file paths using VSCode Agent/chat-titles.txt naming convensions
• Utilize domain-specific templates like Augment Code/claude-4-sonnet-agent-prompts.txt
• Enforce VSCode Agent/gpt-5-mini.txt quality gate procedures for code verification

Future Directions

• Develop scenario-specific prompt templates for autonomous driving applications
• Enhance multi-agent coordination mechanisms from Traycer AI/plan_mode_tools.json
• Implement dynamic parameter adjustment using real-time sensor data inputs