This repository contains AI agent skills designed for Claude, Codex, and other LLM-based systems to support structured legal and engineering tasks.
It is intended both to demonstrate my experience designing AI skills and to provide examples that teams can reference when building AI-enabled legal tools, developer assistants, or enterprise AI systems that require repeatable, well-structured workflows.
AI Skills are structured instruction modules that guide LLMs to perform specialized tasks consistently. Skills may include persistent memory, sub-agents, continuous file system access, and MCP tool integrations when necessary.
Instead of relying on a single prompt, a skill defines AI's:
- Role
- Workflow
- Expected Inputs
- Constraints and Guardrails
- Output Format
This helps produce more reliable results than ad hoc prompting.
Skills are compatible with:
- Anthropic Claude Skills: https://code.claude.com/docs/en/skills
- For instructions to use skills in Claude, see https://support.claude.com/en/articles/12512180-use-skills-in-claude
- OpenAI Codex Skills:https://developers.openai.com/codex/skills
- For instructions to use skills in Codex, see https://developers.openai.com/codex/skills
- Custom AI agents and internal AI tools
I make no representations or warranties regarding the accuracy, completeness, or fitness of these skills for any purpose, and I disclaim liability for any loss or damage resulting from their use.
The legal skills in this repository are not legal advice. They are intended as examples and workflow assets, not as a substitute for a qualified legal professional.
Before using any skill, make sure it is reviewed in consideration of your organization’s confidentiality requirements, professional responsibility obligations, privacy and security standards, and applicable law. Always check AI's work.