Feat: Add agent skills

[wangxingjun778]

1. Add core pipeline for running agent skills
2. Add docs
3. Add examples of skill and data

[gemini-code-assist]

Summary of Changes

Hello @wangxingjun778, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces a robust and modular framework for integrating agent skills into the ms-agent system. It establishes a multi-level progressive context loading mechanism for efficient skill discovery and execution, alongside a structured approach for defining, managing, and utilizing specialized agent capabilities. The changes aim to significantly enhance the agent's ability to perform complex tasks by leveraging a diverse set of tools and knowledge.

Highlights

• Core Agent Skill Pipeline: Introduced a new ms_agent/skill directory containing the foundational pipeline for running agent skills, enabling dynamic skill discovery and execution.
• Skill Management Components: Added dedicated modules for skill loading (loader.py), retrieval (retrieve.py), schema definition (schema.py), and prompt management (prompts.py), providing a structured approach to skill integration.
• AgentSkill Class: Implemented the AgentSkill class, which orchestrates the entire skill execution process, including LLM interaction, code analysis, and script execution, with support for sandbox environments.
• Documentation and Examples: Included comprehensive documentation (projects/agent_skills/README.md) and practical examples of agent skills, such as 'algorithmic-art' and 'pdf' manipulation, to guide usage and demonstrate capabilities.
• PDF Manipulation Tools: Provided a robust set of scripts and documentation within the 'pdf' skill for handling various PDF operations, including form filling (both fillable and non-fillable), text extraction, and image conversion.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request introduces a significant new feature: agent skills. The implementation is comprehensive, including a core pipeline for running skills, loading mechanisms, retrieval, and execution. The code is well-structured with good separation of concerns into different modules like loader, retriever, and schema. The addition of example skills and documentation is also very helpful.

My review focuses on improving robustness, fixing a few bugs, and enhancing maintainability. Key areas for improvement include safer command execution, correcting a bug in skill reloading, and avoiding side effects during context building. Overall, this is a solid foundation for the agent skills feature.