SkillAgentExample

SkillAgentExample is a Python-based agent framework that makes Claude Skills (SKILL.md files) portable across different LLMs using LangChain, with current support for Gemini 2.5 Flash via Vertex AI. The agent exposes three core tools: load_skill to fetch a skill's main SKILL.md instructions, read_skill_file to pull in supplementary reference files like forms.md or reference.md, and run_shell to execute arbitrary shell commands including Python scripts. When a user asks about a domain-specific task such as PDF manipulation, the LLM autonomously decides to load the relevant skill, reads any required reference files, and then executes the appropriate code. This pattern solves the problem of tying specialized agent knowledge to a single LLM provider, letting teams reuse Claude-authored skills in Gemini or other LangChain-compatible models. Developers building multi-LLM pipelines, document automation tools, or skill-driven agents will find this a practical reference implementation.

Install dependencies with `uv sync`, then authenticate with Vertex AI using `gcloud auth application-default login` or a service account key. Edit the query in main.py and run the agent with `python main.py`. The agent will automatically load the relevant SKILL.md files and execute the required code based on your query.