Prompting technique

Plan-Execute Prompting

Learning to Plan & Reason for Evaluation with Thinking-LLM-as-a-Judge, arXiv:2501.18099 (2025) · read the paper

What it does

A two-stage prompting technique: first the LLM creates a detailed plan for solving the task with criteria and structure, then executes the task strictly according to this plan. Dividing into planning and execution phases makes reasoning more structured, reliable, and reduces the probability of errors.

Task typemulti step
Shapechain of thought
Efforttwo or three steps
Addressesthe same prompt giving different answers; answers that are close but wrong; the model losing what was agreed earlier
Published2025-01-30
AuthorsSwarnadeep Saha, Xian Li, Marjan Ghazvininejad et al.

When it helps

Apply this technique for solving complex tasks requiring a structured approach and careful analysis. Especially effective when evaluating content quality, compiling analytical reports, writing large texts, developing business plans, or making multi-factor decisions. Use when reasoning transparency is important and when you need to avoid missing important aspects of a task. The method helps when standard prompts give unpredictable results or the model misses critical details

A prompt you can paste

Generic by design: it applies the technique without knowing your task. Adapt the marked parts.

Prompt
You are an expert in evaluating complex requests and ensuring thoroughness. Your task is to generate a reusable prompt template that guides an LLM through a two-stage process: planning and execution, for any given task.

**### Stage 1: Plan the Execution**

You will first devise a detailed plan for tackling the user's request. This plan must outline the necessary steps, criteria for success, and the structure of the final output. Specifically, for the user's task: "[INSERT USER TASK HERE]", your plan should include:

1.  **Decomposition of the Task:** Break down "[INSERT USER TASK HERE]" into its core components or sub-tasks.
2.  **Identification of Key Criteria:** Define the essential criteria for evaluating the success of the task. These should be measurable and specific.
3.  **Output Structure:** Specify the format, sections, and any specific elements that the final output must contain.
4.  **Potential Challenges & Mitigations:** Identify any foreseeable difficulties in executing the task and propose strategies to overcome them.
5.  **[ADDITIONAL PLANNING ELEMENTS SPECIFIC TO THE USER'S TASK IF NECESSARY]**

**### Stage 2: Execute the Plan**

Now, strictly adhering to the plan you just created in Stage 1, execute the user's task. Ensure that every step outlined in the plan is followed precisely, and the final output matches the specified structure and criteria.

**User Task:** "[INSERT USER TASK HERE]"
**User Context/Files:** "[INSERT CONTEXT OR REFERENCE TO USER FILES HERE]"
**Specific Constraints/Requirements:** "[INSERT ANY SPECIFIC CONSTRAINTS OR REQUIREMENTS HERE]"

If this one does not fit, the two closest alternatives in the corpus are Debugging Prompts Framework and MTR (Multi-Turn Reasoning) Framework, which target the same failure from a different angle.

Worked example

The same technique applied to a concrete job: turn a vague goal into an ordered plan with checkpoints. Use it as the pattern for your own case rather than as a finished artefact.

Worked example
You are an AI assistant tasked with transforming a vague goal into an ordered plan with checkpoints, following the "Plan-Execute Prompting" technique.

Act strictly in two stages:

### Stage 1: Develop a Plan for the Goal

Your plan should outline the steps required to achieve the user's goal. For this task, the goal is: "Turn a vague goal into an ordered plan with checkpoints."

The plan must include the following mandatory sections:
1.  **Goal Clarification:** How to identify and refine the vague goal into a specific, actionable objective.
2.  **Decomposition Strategy:** Methods for breaking down the clarified goal into smaller, manageable sub-tasks.
3.  **Checkpoint Definition:** Criteria and methods for defining checkpoints to monitor progress and ensure alignment with the main goal.
4.  **Execution Framework:** How to structure the execution of sub-tasks and manage progress against checkpoints.
5.  **Review and Iteration:** Process for reviewing outcomes, identifying deviations, and iterating on the plan if necessary.

### Stage 2: Execute the Plan to Produce the Final Output

Now, strictly following the plan you just created in Stage 1, generate the final output. This output should be a structured, ordered plan that clearly details how to turn any vague goal into an actionable plan with checkpoints. Ensure the final output is comprehensive, easy to understand, and directly addresses the initial task.

Get this written for your actual task

Paste what you are trying to do and the corpus will be matched against it directly. Free, no account, about ten seconds.

Free · no signup · ~10s
0.00match confidence
single retrieval pass
Prompt for your task

      

That number is low on purpose, and it is real. It is the raw similarity of one retrieval pass: no specialist read the paper, no judge compared anything, the first plausible match won.

6,235techniques in the corpus
one of which is this page

Picking the right one for a specific task is the work, and it is the work GetDecision does.

This pageone technique, generic prompt
What you just ranone technique matched to your wording, nothing verified
Full runten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history

See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.

Run the full analysis

Related techniques

Debugging Prompts FrameworkA systematic framework for diagnosing and preventing typical errors of language models in instruction understa…MTR (Multi-Turn Reasoning) FrameworkA method of structuring multi-step interaction with an LLM through a "game with rules" format. Includes four b…StateAct (Self-Prompting + Chain-of-States)StateAct improves LLM agent performance through two mechanisms: continuous reminder of the end goal (self-prom…Dynamic Enhancement Chain (DEC)A method that breaks down a complex question into a chain of simple sub-questions solved sequentially. Each st…

All techniques · Failure modes and fixes