Prompt Engineering

Establish Domain Knowledge for AI

AI knows general programming but not your domain. Establish how your specific world works — its rules, its vocabulary, its non-negotiables — so it reasons inside it.

Open in Project Context Builder

Overview

A model can write flawless Python and still misunderstand your data pipeline, because the hard part isn't the language, it's the domain. This setup establishes the domain for Quarry, a batch data pipeline feeding a warehouse. The facts that matter aren't syntax — they're that runs are idempotent, aggregates use UTC day boundaries, and bad rows are quarantined not dropped. The glossary defines run, backfill, DAG, and partition; the never-assume list bars trusting input data. The profile gives the AI the mental model of the domain, not just the code.

How to use this resource

State the domain rules

Idempotent runs, UTC boundaries, quarantine bad rows — the laws of your world.
Define the vocabulary

Run, backfill, partition: the words that carry domain meaning.
Give the AI the model

It reasons inside your domain instead of applying generic defaults.

Why This Works

The domain, not the language, is what general-purpose models get wrong
Always-remember rules encode the invariants the system depends on
Never-assume rules stop the model trusting things your domain never trusts

Best for

Data, fintech, and other rule-heavy domains
Systems with strong invariants
Projects where the language is easy but the domain is hard

Not for

Explaining a specific algorithm — use the Code Explanation Prompt
A behavior persona for one task — use the System Prompt Generator

FAQ

How do I give AI my project's domain rules instead of just the code

The profile captures domain facts in named sections: Always Remember holds invariants like UTC day boundaries and idempotent runs, and the Glossary defines terms like run, backfill, and partition. Project Context Builder generates it; you save it at your repo root as .cursorrules so Cursor loads it for every request and each teammate gets the same context.

How does the profile stop AI from trusting things my system shouldn't

The Never Assume section encodes exactly that: do not assume input data is clean, validate and quarantine bad rows; reference the migrations before writing a query; and when a fact is missing from the profile, ask rather than invent it. These bars stop the model applying generic defaults where the Quarry pipeline has strict non-negotiables.

Should I use this to give the AI a persona or task behavior

No, this is project knowledge, not behavior. The AI Usage Notes state the profile is durable project identity and point you to the System Prompt Generator for a task persona with role, tone, and escalation rules, and to the Context Handoff Builder for carrying in-progress work state. This one establishes the domain's facts and vocabulary only.

Customize This Resource

Opens this project in Project Context Builder. Build to get the full context profile — then adjust the mode, stack, glossary, and output target.

Open in Project Context Builder

Prompt Template

Copy it as-is, or use Open in Project Context Builder to load it pre-filled and customize it with your own context.

# Project Context: Quarry

> Save this at your repo root as `.cursorrules` (or `.cursor/rules/project.mdc`) so Cursor loads it for every request.

## Project Overview
Quarry is a batch data pipeline that ingests raw event logs, cleans them, and loads daily aggregates into the warehouse for the analytics team.

## Core Facts
- Project type: Software Project
- Primary stack: PostgreSQL, SQLAlchemy, Python, Docker, AWS, PyTest
- This is a software codebase — correctness, readability, and maintainability come before cleverness.

## Tech Stack
- PostgreSQL
- SQLAlchemy
- Python
- Docker
- AWS
- PyTest

## Team Conventions
- Every transform is pure and unit-tested on a fixed sample.
- Pipelines are idempotent — re-running a day produces the same result.
- Schema changes go through a migration and a backfill plan.
- Define models declaratively and use sessions correctly; manage migrations with Alembic. (SQLAlchemy)
- Follow PEP 8 and add type hints on public functions. (Python)
- Keep the app twelve-factor — config via environment, no host-specific assumptions in the image. (Docker)
- Use the platform's managed services and least-privilege IAM; never hard-code credentials. (AWS)
- Use fixtures and parametrize for coverage; keep tests deterministic. (PyTest)

## Naming Rules
- Database: snake_case for tables and columns; plural table names. (PostgreSQL)
- snake_case for functions and variables, PascalCase for classes, UPPER_SNAKE for constants (PEP 8). (Python)

## Architecture Principles
- Keep modules focused — a file or function should do one thing well.
- Prefer clarity over abstraction; introduce indirection only when duplication has actually appeared.
- Changes should come with the test that proves them.

## Glossary
- **Run** — one execution of the pipeline for a given date partition.
- **Backfill** — reprocessing historical partitions after a logic change.
- **DAG** _(abbreviation)_ — the directed graph of pipeline steps.
- **Partition** — one day's slice of data.

## Always Remember
- All aggregates are computed in UTC day boundaries.
- This profile is the source of truth for project facts — prefer it over general assumptions.

## Never Assume
- Do not assume input data is clean — validate and quarantine bad rows.
- Do not assume the schema — reference the migrations or schema file before writing a query.
- Do not assume newer syntax is available — confirm the target Python version.
- Do not hard-code secrets or region assumptions — use the configured environment.
- Do not assume how a module is used elsewhere — check the call sites before changing a signature.
- When a fact is missing from this profile, ask — do not invent it.

## AI Usage Notes
- Cursor reads project rules from the repo, so this travels with the code and every teammate gets the same context.
- This profile is project KNOWLEDGE, not behavior — for a task persona (role, tone, escalation rules), build a System Prompt in the System Prompt Generator.
- It is durable project identity, not work state — to continue an in-progress task in a new chat, use the Context Handoff Builder.
- Profile completeness: 100/100 (Complete) — bracketed gaps, if any, mark missing info that was never invented.

More resources from Project Context Builder

Resource

Build Project Memory for AI

Stop re-explaining your project in every new chat. Capture it once as a context profile the AI keeps across every conversation — stack-inferred conventions and all.

Prompt Engineering

Resource

Create an AI Project Profile

A reusable profile that tells AI everything constant about your project — type, stack, conventions, glossary, architecture principles, and the rules it must never guess.

Prompt Engineering

Resource

AI Onboarding Context for an Existing Codebase

Onboard AI to a codebase the way you'd onboard a new hire: what it is, how it's built, the rules that aren't written down, and the things never to assume.

Engineering

Resources that pair well

Resource

AI Session Handoff — Shift Change for Working Sessions

End a working session like a shift change, not an abandonment: state captured, decisions logged, next step named — ready for the next session to pick up.

Prompt Engineering

Resource

Bug Triage Assistant

Convert scattered bug notes, Slack messages, or user complaints into structured engineering tasks with reproduction steps, severity, and root cause hypothesis.

Engineering

Resource

Convert a Conversation to a Prompt — the Full Anatomy

What a converged conversation contains: a rejected voice, an accepted framing, a hard constraint, a format decision, and a final-version moment — each mapped to its prompt section.

Prompt Engineering

Related tools

Tool

Project Context Builder

Define what AI should permanently know about your project — stack conventions, glossary, and never-assume rules in a profile you paste once.

Context Tools

Tip: Save time by exploring related resources and tools that integrate with this resource.