The prompt library
we actually use
No prompt-marketplace bloat. Each prompt lists the use case it solves and the audience it is tuned for, across engineering, product, security, governance, and leadership.
Architecture review: second opinion
Sanity-check an in-progress design doc against unstated risks before commit.
Threat model an LLM feature
Run a 10-minute threat model for any feature that calls an LLM.
Founder: weekly priority cut
Convert a messy backlog into the one thing that moves the needle this week.
Prompt regression eval skeleton
Generate an eval harness for a prompt you're about to change in production.
EU AI Act: risk-tier classifier
Classify an AI system under the EU AI Act's risk tiers and list the obligations that follow.
Internal AI usage policy: one-pager
Draft a plain-language acceptable-use policy for AI tools your staff can actually follow.
Role-based AI literacy quiz
Generate a short, role-specific quiz to gauge a team's real AI fluency before training.
Explain an AI concept to a skeptic
Turn a technical AI concept into a crisp explanation a doubtful stakeholder will accept.
AI use-case prioritizer
Score and sequence a list of candidate AI use cases by value, feasibility, and risk.
Build vs. buy vs. wait memo
Get a one-page decision memo for an AI capability you're deciding how to source.
AI initiative: back-of-envelope ROI
Sanity-check the payback on an AI initiative before you commit budget.
Data readiness gap scan
Find the data gaps that would block an AI use case before you start building it.
AI vendor / model evaluation rubric
Build a scoring rubric to compare AI vendors or models on more than the demo.
RAG retrieval debugger
Diagnose why a RAG system is returning weak or wrong context.
Agent tool / function spec writer
Write a clean, model-friendly tool definition for an LLM agent to call.
Data pipeline design review
Pressure-test an ETL / data-pipeline design before you build it.
Cloud cost guardrails for AI workloads
Put guardrails on AI/inference spend before the bill surprises you.
UX patterns for AI uncertainty
Design how your interface should behave when the AI is unsure, slow, or wrong.
Prompt-injection test suite generator
Generate adversarial test cases to probe an LLM feature for prompt injection.
LLM feature incident runbook
Draft an on-call runbook for when an AI feature misbehaves in production.
Model drift & quality monitoring plan
Set up monitoring that catches quality regressions and drift after launch.
AI feature handover checklist
Make sure a team can fully own an AI feature after the builders step away: the keys-in-hand checklist.
On-call triage: my AI feature is acting up
A fast triage walkthrough when an LLM feature starts misbehaving and you're on call.
PRD from a one-line idea
Turn a one-line feature idea into a tight, reviewable PRD a team can argue with.
Heuristic UX review of a flow
Run a fast usability-heuristics pass on a screen or flow before you ship it.
Blameless incident postmortem
Turn raw incident notes into a blameless postmortem with real action items.
On-call runbook for a service
Generate a practical on-call runbook for a service from a short description.
Exec-ready status update
Compress a messy project update into a crisp status an exec reads in 30 seconds.
Weekly 1:1 prep and coaching
Prep a high-signal 1:1 agenda from recent context, not a status readout.
Source triage for a research question
Triage a pile of sources into what to read first and what to skip, against a real question.
Three-touch cold outbound sequence
Draft a 3-email cold sequence that earns a reply without sounding like a template.
Pricing and packaging stress test
Pressure-test a pricing and packaging plan before you put it in front of customers.
Customer discovery interview script
Generate a Mom-Test-style discovery script that gets facts about the past, not opinions about your idea.
SQL explainer and optimizer
Explain a SQL query in plain English and suggest safe optimizations without changing results.
Define one metric precisely
Pin down one metric's exact definition, edge cases, and how it can be gamed before anyone reports it.
Pull-request review pass
Review a PR diff against a senior-reviewer checklist, sorted by what actually matters.
These are the prompts behind production work. When you want the workflows and apps they sit inside built and handed over, that is Build & Run.
See Build & Run →