Legal work is pattern recognition, and AI slots in naturally. I’ve sat through enough contract reviews to know where the boring parts eat hours: reading the same indemnity clause for the 200th time, prepping a witness for another deposition, hunting a typo in a clause someone copy-pasted across three files. This pack hits three of those spots. Each prompt hands back something usable, not a lecture. One caveat: AI drafts, but a human lawyer still signs. If you’re using this for real client work, your bar number is doing the heavy lifting, not the model.
[Claude]
You are a commercial contracts reviewer. Given the full text of my SaaS master services agreement, produce a risk redline checklist in a Markdown table. Columns: clause reference, plain-English risk, severity (High/Med/Low), and a suggested redline edit. Flag only terms that deviate from market standard for a $50k annual deal with 12-month auto-renewal. Output the table plus a 5-point negotiation priority ranked by dollar exposure. Do not summarize the contract; audit it.
[ChatGPT]
Act as a litigation prep partner. I defend a warehouse manager in a premises liability deposition next week. Given the complaint, my client’s notes, and the plaintiff’s medical chronology, build a witness prep outline: the 6 factual claims most likely to trip him up, 4 safe ways to answer without over-explaining, and 5 cross-exam questions opposing counsel will likely use. End with a one-page ‘do not say’ sheet he can review tonight and cite which source each item came from.
[Cursor]
You are refactoring a legal document generator codebase. Given the repository at ./contract-forms, find the function that builds the indemnification clause for our non-disclosure templates. It currently uses string concatenation and repeats the same boilerplate in 3 files. Refactor it into a single reusable module with: typed parameters for scope and liability cap, unit tests covering the default and capped cases, and inline docs. Output the new file paths and a diff summary. Do not change rendered output.
Want more? Full archive is at aitoolalliance.com/tag/prompt-pack.