Reprompt
Turn a rough idea into a prompt that actually knows your codebase.
The Problem
A good prompt takes context, structure, and specificity most people don't have time to type out by hand every time — so prompts stay vague, and answers stay generic.
Reprompt's core idea is that the gap between 'a rough idea' and 'a genuinely good prompt' is mostly mechanical — context, structure, constraints — and that mechanical gap can be closed automatically. Type a simple instruction, and Reprompt expands it into a structured, XML-tagged prompt (context, instruction, examples, constraints, format) that already accounts for the tech stack it detects in your open project.
Teams can go further by defining shared prompt rules once, in .rpmt or .rprmt files, so everyone on a project is working from the same standards instead of reinventing prompt structure per person. Once a prompt is built, Reprompt can run it directly against Perplexity's Sonar model family — sonar, sonar-pro, sonar-deep-research, sonar-reasoning, sonar-reasoning-pro, and r1-1776 — with adjustable web search context depth, right inside the editor.
Every transformation comes back with rich, copyable markdown and a small set of honest stats — expansion ratio, processing time — so you can see exactly what got added and why, rather than trusting a black box.
Who It's For
Developers who use AI models daily for coding help and are tired of re-typing the same context — stack, conventions, constraints — into every single prompt.
"I noticed I was typing the same 'by the way, this project uses X, follow Y convention' preamble into almost every prompt. Reprompt automates that preamble so I can just say what I actually want."
— Founder, Naviware
Features
- •One-command prompt transformation with automatic project tech-stack detection
- •Custom prompt rules via .rpmt / .rprmt files — define standards once, apply them across the whole team
- •Executes prompts directly against Perplexity Sonar models: sonar, sonar-pro, sonar-deep-research, sonar-reasoning, sonar-reasoning-pro, and r1-1776
- •Adjustable web search context depth — low, medium, or high — depending on how much grounding a prompt needs
- •Rich markdown rendering in-editor with copyable code blocks
- •Structured XML-tag output — context, instruction, examples, constraints, and format — for consistent, parseable results
- •Built-in transformation stats: expansion ratio and processing time for every prompt run
Requirements
- A Perplexity Sonar API key
Built With
Ready to try Reprompt?
Install in VS Code