Advanced Claude

I want your help with [task].

Before producing the final answer, build a working understanding of the situation.

Here is the context:

Goal:
[What I am trying to achieve]

Audience:
[Who this is for]

Current situation:
[Relevant background]

Constraints:
[Time, budget, format, policies, limitations, or requirements]

What has already been tried:
[Previous work or failed approaches]

What a strong result should accomplish:
[Definition of success]

First, summarize your understanding of the assignment. Then identify any missing information or assumptions that would materially affect the result. Do not begin the final work until you have completed this assessment.

Do not begin producing the final deliverable yet.

First, create a plan for completing this task well.

Your plan should include:

The major questions that must be answered

The information required

The sequence of work

The decisions that need to be made

The major risks or failure points

The criteria you will use to evaluate the final result

After creating the plan, review it and identify anything missing, redundant, or out of order.

Then present the improved plan and wait for the next instruction.

Task:
[Describe the task]

Before answering my question, evaluate whether I am asking the right question.

Analyze the premise behind my request.

Identify:

Assumptions I am treating as facts

Information that would change the recommendation

Alternative explanations I may be overlooking

Ways my framing may be too narrow

The strongest argument against my current position

Then rewrite the original question into a stronger version.

After that, answer the improved question rather than blindly accepting my original framing.

My question:
[Insert question]

Perform an assumption autopsy on the following plan:

[Insert plan, proposal, idea, or strategy]

Extract every meaningful assumption required for this plan to succeed.

Organize the assumptions into these categories:

Customer assumptions
Market assumptions
Behavior assumptions
Operational assumptions
Financial assumptions
Technical assumptions
Timing assumptions
Organizational assumptions

For each assumption, provide:

The assumption

Why the plan depends on it

Evidence currently available

What would prove it wrong

Consequence if it is wrong

The fastest realistic way to test it

Then rank the assumptions from highest risk to lowest risk using uncertainty and impact.

Do not summarize the plan. Focus on exposing what must be true for it to work.

Review the answer you just produced as a demanding expert editor and critical reviewer.

Do not defend the answer.

Identify:

Generic or predictable sections

Claims that lack evidence

Important nuances that were missed

Weak assumptions

Repetition

Places where the answer avoids making a decision

Advice that sounds correct but is difficult to apply

Anything written to sound intelligent rather than create value

Score the answer from 1 to 10 for:

Accuracy
Specificity
Originality
Practical value
Clarity
Depth

Then rewrite the answer to address every important weakness you found.

Generate four meaningfully different approaches to this problem:

[Describe problem]

The approaches must differ in their underlying strategy, not merely their wording.

For each approach, explain:

Core idea

Why it might work

Required resources

Major risks

Time to meaningful evidence

Best situation for using it

Reason to reject it

Then compare the four approaches against these criteria:

Impact
Speed
Cost
Risk
Reversibility
Evidence required
Operational complexity

Recommend one approach, but also explain which new information would cause you to change the recommendation.

Use the attached reference material as the primary source for this task.

Before answering:

Identify the sections relevant to my request

Separate explicit facts from interpretations

Note any contradictions or gaps

Preserve the terminology used in the source

Do not add unsupported claims as though they came from the material

When producing the answer:

Use the sources to support specific conclusions.
State when the available material does not answer a question.
Label any recommendation that requires your own inference.
Do not replace the source's framing with generic advice.

Task:
[Insert task]

Evaluate the following decision using a five-person advisory board.

Decision:
[Describe the decision]

The advisors are:

The Strategist
Focuses on positioning, competitive advantage, and long-term consequences.

The Operator
Focuses on execution, process, staffing, dependencies, and operational risk.

The Customer Advocate
Focuses on customer behavior, pain, adoption, trust, and willingness to pay.

The Skeptic
Looks for weak assumptions, hidden costs, downside risk, and reasons the plan may fail.

The Financial Advisor
Focuses on cost, cash flow, return, opportunity cost, and measurable outcomes.

Have each advisor provide:

Their recommendation

Their reasoning

Their biggest concern

The evidence they would want

One action they would take next

Then synthesize the discussion.

Identify where the advisors agree, where they disagree, and which disagreement matters most.

Finish with one recommendation, the conditions under which it is valid, and the next three actions.

Before completing this task, define the evaluation rubric you will use to judge the result.

Task:
[Insert task]

Create six to eight criteria for excellent work.

For each criterion, explain:

What strong performance looks like

What weak performance looks like

Common failure modes

How much weight the criterion should receive

Then complete the task.

After completing it, score the result using the rubric.

Revise any section that scores below 8 out of 10.

Present only the revised final version and a brief score summary.

Turn the following recurring task into a reusable Claude workflow:

[Describe recurring task]

The workflow should include:

Required inputs

Questions Claude should ask before beginning

Step-by-step process

Decision points

Quality checks

Common failure modes

Final output structure

Instructions for revising weak output

Then create a master prompt I can reuse every time I perform this task.

The master prompt should:

Use placeholders for changing information
Preserve the full process
Ask only necessary questions
Separate analysis from final output
Include a self-review step
Produce consistent formatting
Avoid generic recommendations

After writing the master prompt, test it against one realistic example and identify any weaknesses. Then provide the improved final version.