liveBook · Manning

Prompt

As explained in Claude at Work

A prompt is an input message that specifies a question, instruction, or requested task for an AI system.

A prompt is a question or request sent to an AI system. In a conversation with Claude, it can be accompanied by files or guidelines; the response may include an answer or a requested deliverable. Claude may also use tools while working on the request. section 1.1

An opening prompt sets the direction, but it does not need to be perfect. A clear, brief and specific request can reduce clarification later. There is no fixed formula: include the information relevant to the task. section 2.4 section 2.9

For a useful first prompt, consider four elements:

  • Goal: say what you want done and which deliverable you need. section 2.4
  • Context: supply relevant assumptions, such as the existing infrastructure, preferred stack, scale and team constraints. section 2.4
  • Constraints: identify limits, such as security restrictions or budget. section 2.4
  • Format: specify the structure, sections or output format, or provide a template. section 2.4

Develop an architecture design

Attach a rough architecture sketch and start with “What do you make of this?” Then ask what is missing, discuss technology options and data flows, and request a design document. Review the draft and use a more specific follow-up to change its summary, diagram and section order. A broad opening can start exploration; later prompts can direct revisions. section 1.5.1

Treat the response as work to review, not an automatic answer. If something is wrong, state the correction and add evidence such as documentation, logs or screenshots. Continue with feedback and examples when useful; if the conversation is going sideways, start a new one with an improved prompt. section 2.7.2 section 2.9

When a design is ready for review, ask whether it has enough detail for the next level of component designs. Use the answer alongside your own judgment to decide what still needs work. section 1.5.2 section 2.9

How Claude at Work covers it

How deep it goes
Explains opening-prompt structure and follow-up corrections, without prescribing a fixed formula.
Examples
Uses an architecture sketch and design-document revisions, plus a cloud-migration question, to illustrate requests.
What you can do after
Apply it via Claude to frame a task, review the response and refine the output through follow-up requests.
How it connects
Connects prompts with conversation, context, constraints, output formats and artifacts.

Also explained in

Compare how 64 books explain prompt →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Claude at Work. The section links point to the parts of the book it draws on.