Build a deliberate code analysis packet by selecting the smallest sufficient set of repository files, stating the question and constraints, and giving a reasoning model exact paths and contents without hidden context or irrelevant generated code.
Source episode · 26:45
How I AI: Ryan Carson's 3-Step Playbook for Structured AI Development in Cursorwith Claire Vo
Ryan uses Repo Prompt to select the exact application, component, library, schema, and script files needed for a maintainability question, removes generated and irrelevant content until the token set is manageable, adds an architect prompt, and sends the clearly demarcated file context to a reasoning model.
A reproducible prompt packet containing repository revision, selected file manifest, exact code context, task, constraints, and requested response format, plus findings that cite files and remain hypotheses until verified.
Anthropic AI assistant
Google AI assistant
Step by step
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
4 steps
Start from the question and select its entry point, direct callers, types, schemas, configuration, tests, and one representative implementation. Watch the token count and remove generated, vendored, or unrelated files.
Write a precise request with the desired outcome, constraints, non goals, environment, and response format. Add an architect or reviewer perspective only when it improves the requested analysis.
Analyze the selected repository context at commit [commit] for [question]. Constraints: [constraints]. Cite exact files and symbols. Return current behavior, dependency map, risks, missing context, and two options with tradeoffs. Do not invent unprovided code or modify files.
Generate the packet and inspect the manifest before copying it. Confirm file paths are clear, content is complete, secrets are absent, and the task appears after the context in an unambiguous format.
Send the packet to the approved model and treat its answer as an analysis proposal. Open every cited location, request any missing context, reproduce claimed behavior, and run focused tests before implementing a recommendation.
Turn an idea into a PRD, user stories, and a plan.
After the steps
How to recover when the loop fails and where human judgment helps.
Recover
Keep building
Design a weekly meta-workflow that analyzes your interactions with your AI assistant to make it smarter. This system learns your writing style from edits, suggests new automations for repetitive tasks, and helps you filter valuable AI techniques from hype.
Create a daily AI workflow that summarizes yesterday's meetings, identifies urgent tasks, and proactively learns internal company terms it doesn't understand. This keeps your AI assistant's knowledge current and highly relevant to your work.
Build a recurring AI task that automatically gathers updates from your calendar, Slack, and notes to suggest weekly priorities and prepare you for upcoming meetings. Start every Monday with a clear, comprehensive plan without the manual effort.
Join 100,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.