Claude Skills for Lawyers

NDA Review Against Your Playbook compares a counterparty’s NDA with your client’s playbook, clause by clause. An associate or in-house counsel pastes the NDA clauses and the client’s playbook positions. The skill returns an issue table, risk levels, proposed redline language, and questions for the supervising attorney. The output is a draft for the responsible attorney to check. For prompts, see our prompt library for lawyers.

To install in Claude Code, download the ZIP, unpack it into ~/.claude/skills/ (or .claude/skills/ inside your project), and restart Claude Code. The skill loads when your request matches its description. If you don’t use a terminal, open a Claude Project, go to Project instructions, paste the SKILL.md text, and save. Every chat in that project follows those instructions.

The models listed on the page were tested on these dates: Claude Opus 5.5 and Claude Sonnet 5 on 2026-10-19, and Claude Fable 5.1 on 2026-10-20. Each model ran the skill 3 times on a fictional input written with its rubric before the skill existed. A model from another family, gpt-6-astra, graded every run. See the methodology for the testing setup.

This skill is written for US practice. It refers to ABA Formal Opinion 512 (July 2024), from the ABA Standing Committee on Ethics and Professional Responsibility, on competence, confidentiality, supervision, and fees when lawyers use generative AI. State bar rules may add duties, so the supervising attorney checks the rules in their state. The issue table and proposed redlines are suggestions for attorney review; the supervising attorney also receives questions alongside them, rather than only proposed wording for the NDA.

The skills are free to use, adapt, and share, including commercially, under CC BY 4.0 with attribution.