CoreWise Academy

Layer II · THE LOOM

Prompting & Context

Getting the right material in front of the model: instructions, examples, and retrieved context, arranged so what matters stands out.

Broad

1 guide
  1. Nº 014 Brief the model like a brilliant new hire 6 min

    Claude is a brilliant new hire with no context on your job. Brief it like one: explicit instructions with motivation, examples, XML tags, roles, and context ordering.

Practitioner

6 guides
  1. Nº 015 Stop shouting at the model 5 min

    The newest Claude models are highly responsive to system prompts, so shouting backfires: soften trigger language, migrate deprecated levers, keep the model in bounds.

  2. Nº 011 Thinking on a budget 9 min

    Reasoning tokens are billed at the output rate and re-read every turn after. How to compress what a model says without compressing what it knows.

  3. Nº 024 Sharpen the prompt or brainstorm first 5 min

    Two ways to start an AI task: deepen a rough prompt when you know the goal, or brainstorm first when the goal or the tradeoffs are still open. A quick test for which one you are in.

  4. Nº 025 Power start sessions with /enhance-prompt 6 min

    An enhance-prompt skill turns a rough request into the full prompt the model will actually run, and shows it to you first. Reading that rewrite catches misunderstandings before any work starts.

  5. Nº 030 Trim the instructions your model outgrew 6 min

    Anthropic removed over 80 percent of Claude Code's system prompt for the Claude 5 models and measured no loss on its coding evaluations. How to give your own instruction files the same trim: goals instead of rule lists, context loaded on demand, and a /doctor audit.

  6. Nº 048 Take a problem apart before AI solves it 6 min

    A language model reaches for the most familiar answer, and the better it sounds the less you check. Four staged prompts force the opposite: break the problem into parts, sort facts from assumptions, recombine the checked parts, and design cheap tests. You keep the judgment.

Deep

1 guide
  1. Nº 016 Guardrails for hours-long agent runs 6 min

    Guardrails are the support structure that keeps an hours-long agent run honest: an audit instruction, a memory that survives resets, and fresh-eyed verifiers.

Field notes on this layer

Prompting is the craft of briefing a model, and context is the material you brief it with. This layer treats the two as one discipline: what you say, what you attach, and how you arrange both so the model attends to what matters instead of guessing.

The prompting guides run from the fundamentals that hold across every current model to the context budgets that decide what an agent session costs. Reading the whole layer takes 26 minutes; the habits it replaces waste far more than that every week.

LevelGuidesReading
Broad 1 6 minutes
Practitioner 6 37 minutes
Deep 1 6 minutes
Whole layer 8 49 minutes

Quotes from the guides

“A colleague who is confused, Claude will be too.”

Brief the model like a brilliant new hire

“You are steering a model that already listens.”

Stop shouting at the model

“Reasoning tokens are billed twice.”

Thinking on a budget

Primary sources this layer teaches from

FAQ · the Prompting & Context layer

What does the Prompting & Context layer cover?

Getting the right material in front of the model: instructions, examples, and retrieved context, arranged so what matters stands out. Guides are ordered Broad, then Practitioner, then Deep, and each one lists its prerequisites.

How long does this layer take to read?

The layer currently holds 8 guides, about 49 minutes of reading in total. Each guide also ends with a short exercise you can run the same day.

Do I need to read the guides in order?

No. Levels run Broad to Deep and prerequisites are marked on every guide, so you can enter anywhere and follow the marked links back when a guide assumes something you have not read.