$ cat ~/work/caseprompts.md
CasePrompts
Prompt like you're assigning the work — not Googling it.
$ cat ~/caseprompts/problem.md
The problem
Most people prompt an AI the way they'd type into a search box — a few words, a vague ask — and get back something confident, fluent, and unsourced. In litigation, an answer that isn't tied to the record isn't just unhelpful; it's dangerous. The model will happily fill a gap with something that sounds right and isn't in the transcript at all.
The fix isn't a better model. It's a better prompt — one that assigns the work with the same precision you'd give a sharp associate.
$ cat ~/caseprompts/cite-method.md
The CITE method
CasePrompts is built on CITE — Context, Instructions, Template, Evidence. You give the model the context of the matter, explicit instructions for the task, the exact template you want the output in, and the evidence it's allowed to work from. Structure replaces guesswork.
It turns prompting from a wish into an assignment: here's the record, here's the job, here's the format, and here are the rules.
$ cat ~/caseprompts/discipline.md
The page:line discipline
Every CasePrompts output cites page:line. If a claim isn't in the record, the prompt tells the model to say so instead of inferring. That single rule is the difference between a tool a litigator can trust and one they have to double-check line by line.
It comes from the right place: these are built by people who produce the record, where citing the source isn't a nicety — it's the whole job.
$ ls ~/caseprompts/library
What's in the library
A free, growing set of copy-ready prompts for real litigation work — deposition summaries, medical-record review, discovery, expert prep — each one ready to paste into the AI of your choice. No signup, no lock-in.
It's in preview and building in public under DepoStack, with new templates added as the method gets sharpened against real work.