Step01
Create an Agent Instruction File
Create a versioned Markdown instruction file under the repository’s agent or instructions directory. Define its scope, intended document type, output folder, and filename convention.
IntermediateProduct
How to Create a Reusable AI Agent in Cursor for Consistent Document Generation
Encode a compact document standard in a repository instruction file, invoke it with a specific brief, and improve the reusable instructions from the rendered output rather than rewriting every document by hand.
Source episode · 27:00
How I AI: A Beginner's Guide to Coding with AI—Build Your First Project Hubwith Claire Vo
Claire creates a reusable PRD instruction file in Cursor, narrows it for a functional solo-project format, invokes it to write a neighborhood task-scheduler PRD, and checks the Markdown in her hub.
Before you start
What you need
- A repository with a Markdown documents directory
- One example of a strong document
- A clear audience, purpose, and required sections
- A file naming and output-location convention
What you’ll make
A reusable instruction file that produces concise, consistent Markdown documents in the correct directory without inventing product facts or overwriting existing work.
Tools used
- CursorVisitCursor
AI-first code editor
Step by step
The workflow
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
4 steps
Step02
Generate the Agent's Instructions
Ask Cursor to populate it with a compact template, section rules, evidence and assumption handling, length guidance, and a pre-write collision check.
Example prompt
Fill this instruction file for generating concise PRDs in @docs. Require: problem, target user, evidence, goals and measures, user journey, functional requirements, non-goals, constraints, technical notes, decisions, assumptions, open questions, and acceptance criteria. Do not invent facts. Use a slug filename, check for collisions, and report the output path. Keep the instruction file maintainable and under [limit].
Step03
Use Your Custom Agent to Write a Document
Invoke the instruction file from a new prompt and provide only the product-specific brief, evidence, and constraints. Ask the agent to label missing information instead of guessing.
Example prompt
Using @agents/prd.md, write a PRD for [product]. User: [user]. Problem evidence: [evidence]. Core journey: [journey]. Constraints: [constraints]. Known decisions: [decisions]. Save it under @docs and do not overwrite an existing file.
Step04
Verify the Output
Open the generated Markdown in the app and inspect structure, length, assumptions, requirements, and rendering. Correct the document, then update the instruction file when the correction should apply to future PRDs.
What good looks like
- Generated files follow the required structure and naming rules.
- Inputs, assumptions, decisions, requirements, and open questions remain distinct.
- The instruction file is short enough to maintain and version with the repository.
- A second document can be generated without copying the full template into the prompt.
Build your next product with ChatPRD
Turn an idea into a PRD, user stories, and a plan.
After the steps
Runbook notes
How to recover when the loop fails and where human judgment helps.
Recover
If it goes sideways
- The instructions produce generic or bloated documents
- Add a concrete example, section budgets, required decision fields, and explicit exclusions.
- The agent fills missing product information with confident assumptions
- Require unknown, assumption, and open-question labels, and forbid invented users, metrics, or constraints.
- The generated file lands in the wrong directory or overwrites another document
- Define a slug-based filename, require a collision check, and print the created path.
- People change generated documents but never improve the reusable instructions
- When the same correction appears twice, update and commit the instruction file with a regression example.
Keep building
Related workflows
advancedEngineeringClaude
How to Implement AI Governance with Context-Aware Project Controls
Create context-aware guardrails for your enterprise AI agent using a 'Project' governance layer. As a project owner, define rules, scope tools, and require human-in-the-loop approval for sensitive actions, ensuring safe and effective AI use.
Open
intermediateOperationsKai (Stripe)
How to Turn a One-Off AI Chat Session into a Reusable Workflow for Your Team
Transform a successful AI chat session into a reusable skill that anyone in your company can use. Package prompts, tools, and code into a standardized workflow, making it easy to share and scale automation across your organization.
Open
beginnerDataKai (Stripe)
How to Create a Custom Data Dashboard On-the-Fly with a Natural Language AI Agent
Use a natural language AI agent to instantly generate and iterate on custom data dashboards. This workflow queries your company's data securely, allowing non-technical users to perform their own 'last mile' data analysis without filing tickets.
Open
Start shipping
better products.
Join 100,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.
Free to start
No credit card
SOC 2 certified
Enterprise ready