AI Systems/AI & reliability

Prompt Engineering

5 micro-topics in AI Systems, each with the evidence that proves you have it and the written reason behind every link. Before you start, it rests on 2 other domains. Downstream, it holds up 3 domains.

5 topics · depth 0–7 of 16 · 4 internal links · 2 in · 3 out

Prerequisites · 2 links in

Before you start

Load-bearing links first. Each one says what in this domain rests on what outside it, and why.

Model Integration

1 link

load-bearing

Write direct instructions rests on Configure a model client

You need something to prompt.

Version Control

1 link

load-bearing

Version prompts as templates rests on Stage and commit changes

Versioning prompts requires version control.

5 topics · depth 0–7

What you will be able to do

In prerequisite order. Each idea names the artifact that closes it, the market demand that put it on the map, and what it stands on.

Depth 0 · meta · scaffolded · M18 §18.2

Recognize obsolete prompting tricks

Knows which techniques stopped paying and why

Evidence: Removes an incantation and shows the score did not drop
Market anchor: AI engineering
Rests on: supporting
Use examples deliberately
You can only retire a technique you have used.

Depth 5 · representational · unassisted · M18 §18.1

Write direct instructions

States the task plainly with the needed constraints

Evidence: A zero-shot instruction produces usable output
Market anchor: Prompt engineering
Rests on: load-bearing
Configure a model client · Model Integration
You need something to prompt.

Depth 6 · conceptual · unassisted · M18 §18.1

Use examples deliberately

Adds examples where format matters and omits them where it does not

Evidence: Justifies including or dropping few-shot examples
Market anchor: Prompt engineering
Rests on: load-bearing
Write direct instructions
Examples supplement an instruction.

Depth 6 · representational · unassisted · M18 §18.3

Structure prompts for reliable parsing

Uses delimiters and format constraints

Evidence: Output parses on the first attempt across a test set
Market anchor: Prompt engineering
Rests on: load-bearing
Write direct instructions
Structure organizes an instruction.

Depth 7 · meta · scaffolded · M18 §18.5

Version prompts as templates

Treats prompts as code with variables and history

Evidence: A prompt change is reviewable in a diff
Market anchor: Prompt engineering · MLOps
Rests on: load-bearing
Structure prompts for reliable parsing
Templates parameterize structured prompts.
load-bearing
Stage and commit changes · Version Control
Versioning prompts requires version control.