Loops & Prompts
The best prompts and agentic loops for founders and builders, by category. A prompt gets one job done well; a loop keeps an agent going until the job is verifiably done.
What makes these different
Fewer, better prompts for real founder and builder work. Each one has the inputs to fill in, a step-by-step method, an output format and a self-check.
How to use one
Copy it, or open it straight in Claude or ChatGPT, then replace the highlighted inputs with your details. Need it to repeat until done? Use a loop.
Built on the best practices of
prompts.chat ↗ The successor to Awesome ChatGPT Prompts: the largest open (CC0) prompt library.
Anthropic prompt library ↗ Anthropic's examples, and its prompting guide on structure, examples and XML tags.
OpenAI Cookbook ↗ Worked examples for reliable outputs, structured data and evals.
Gemini prompt gallery ↗ Google's categorized examples for research, coding and extraction.
Sharp positioning in one pageTop pick
Turns a fuzzy product description into a positioning statement, the alternatives you beat and the proof behind each claim.
Pre-mortem for a big betTop pick
Imagines the bet failed a year from now, then works backward to the causes, early warning signs and what to do now.
Cuts a long wish list down to three measurable outcomes, the bets under each, and what you will not do.
Lays out options, criteria and a recommendation for a decision you keep going back and forth on.
Turns a month of metrics into a clear narrative: what happened, why, and what you are doing about it.
Breaks a problem down to what is actually true, then rebuilds a solution without inherited assumptions.
PRD from messy notesTop pick
Turns call notes, Slack threads and a rough idea into a tight PRD with scope, non-goals and open questions.
User interview synthesisTop pick
Finds the patterns across interview transcripts, with quotes, frequency and what to build or stop.
Scores a list of features on reach, impact, confidence and effort, and explains the top five.
Walks a new user's first ten minutes step by step and finds where they stall, with fixes ranked by effort.
Turns a changelog or list of merged PRs into short release notes written around what users can now do.
Reviews your pricing against value metrics, competitors and buyer psychology, with specific changes.
Finds the shared traits of your best customers and turns them into an ICP you can target and disqualify against.
Landing page copy that convertsTop pick
Writes a full landing page from your positioning: hero, problem, proof, how it works, objections and CTA.
A day-by-day launch plan across your channels, with owners, assets and the one metric for each day.
Turns what you already know into 12 post ideas with hooks, angles and the proof for each.
A brief for a page that ranks in Google and gets cited by AI answers: intent, outline, entities and FAQs.
A fair, specific comparison against one competitor that wins on the criteria buyers actually use.
Cold email that gets repliesTop pick
A short, specific cold email and two follow-ups built around a trigger and one clear ask.
Questions to ask, in order, to qualify a deal and find the real pain in 30 minutes.
The objections you hear most, each with the question behind it and a response that does not sound defensive.
A recap from your call notes that confirms pain, agreed next steps and dates, ready to send.
Maps the buying committee, their priorities and a path to a signed deal at a large account.
Monthly investor updateTop pick
A crisp monthly update: highlights, numbers, lowlights and specific asks your investors can act on.
Pitch deck narrativeTop pick
The story arc for a seed or Series A deck, slide by slide, with the one point each slide must land.
Checks whether a fund fits your round: thesis, check size, recent deals, partner and conflicts.
Drafts clear answers to the questions investors ask in diligence, flagging what needs a document.
Explains each clause in plain English, what is standard, and the questions to ask your lawyer.
Thorough code reviewTop pick
Reviews a diff for bugs, security, performance and readability, ranked by severity with fixes.
Debug with hypothesesTop pick
Turns a bug report into ranked hypotheses, the cheapest test for each, and the fix once confirmed.
Compares technical options against your constraints and writes the ADR with consequences.
Writes focused tests for a function or module, covering edge cases you would otherwise miss.
A step-by-step plan for a schema, framework or data migration with zero downtime and a rollback at each step.
A map of a codebase: entry points, data flow, key modules and where to make a given change.
Write a production system promptTop pick
Drafts a system prompt for an AI feature: role, context, rules, tools, output format and examples.
Build an eval setTop pick
Creates test cases and grading criteria for an LLM feature, so prompt changes are measured, not guessed.
Specifies the tools an agent needs: names, descriptions, parameters and failure modes the model can use well.
Finds whether bad answers come from retrieval, chunking, ranking or generation, with tests for each.
Rewrites a weak prompt with clear structure, examples and an output format, and explains every change.
Market mapTop pick
Maps a market into segments with the leading companies, how they make money and where the white space is.
Breaks down one competitor's product, pricing, go-to-market and weaknesses into what to do about them.
Breaks a claim into checkable parts, weighs the evidence and gives a verdict with confidence.
Analyzes a table or export and tells you what it means, what is noise and what to do.
Summarizes what is known about a technical topic, the main approaches and the open questions.
Edit for clarityTop pick
Tightens your draft without changing your voice: shorter sentences, plain words, the point up front.
Drafts a difficult message (bad news, a no, a firm boundary) that is clear, kind and short.
Writes a launch, funding or milestone announcement for LinkedIn or X that people actually read.
Gives you the key points, decisions and action items from a long doc, contract or transcript.
Turns a transcript or rough notes into decisions, owners, dates and open questions.
Writes a role around outcomes in the first year, not a wish list of requirements.
Designs interview stages, questions and a scorecard so every candidate is judged on the same criteria.
Turns how you do a recurring task into a step-by-step SOP someone else can follow the first time.
Builds a simple monthly budget and runway with three scenarios and the levers that matter.
Compares tools against your requirements with a weighted score and the questions to ask each vendor.