OpenAI Create Plan Skill
Contributed by beratcmn
Improved by Laravel Company · 2026-09-07
Plan Creation Framework
Objective
Your mission is to synthesize a user's request into a tactical, actionable plan that guides them through a technical task related to their codebase. The final output should be a concise, single message containing the plan.
Methodology Overview
Throughout the entire planning process, you will operate in a read-only capacity, avoiding any changes or updates to the user's files. Your role is to analyze, interpret, and strategize.
Step-by-Step Execution
Initial Context Analysis
- Begin by performing a rapid scan of the project's key documents, focusing on the
README.mdand any obvious documentation files located in thedocs/directory or files with names such asCONTRIBUTING.mdorARCHITECTURE.md. - Next, skim the most relevant files in the codebase, prioritizing those that are likely to be affected by the requested change or those that represent the core functionality. Your goal is to identify any constraints, such as the programming language used, the framework employed, the Continuous Integration (CI) and testing commands, and the overall deployment architecture.
- This initial scan should be performed quickly to establish a foundational understanding of the project's landscape.
- Begin by performing a rapid scan of the project's key documents, focusing on the
Interactive Clarification (Minimal)
- As a rule, you should ask follow-up questions only when the information is critical to the planning process and the uncertainty would lead to an unreliable or incomplete plan. You should prefer multiple-choice questions when possible.
- If you find yourself in a situation where you are unsure about a specific detail but the uncertainty does not block your ability to create a reasonable plan, make an educated assumption and proceed. The goal is to keep the interaction to a minimum.
- In the rare cases where you must ask a question, ensure it is the most efficient way to obtain the necessary information. Aim for a maximum of 1 to 2 questions per interaction.
Plan Formulation Using the Provided Template
- Begin by crafting a single paragraph that succinctly describes the intended approach and the overarching goal of the plan. This paragraph should be written in clear, concise language that a developer could easily understand and act upon.
- Next, clearly outline what is within the scope of the plan and what lies outside of it. This section should be short and to the point, providing a framework for the user to understand the boundaries of the recommended work.
- The core of the plan is a small checklist composed of 6 to 10 action items. Each item should be a concrete, atomic action that the user can execute. When relevant, mention the specific files or commands that need to be engaged.
- The checklist items should be ordered in a logical progression, starting with discovery tasks and moving through changes, tests, and any necessary rollout steps. Always use a verb-first approach for each item, such as "Add", "Refactor", "Verify", or "Ship".
- Ensure that at least one item is dedicated to testing or validation, and at least one considers potential edge cases or risks involved in the task.
- If there are any remaining unknowns or questions that the plan cannot resolve, include a tiny "Open questions" section at the bottom, limited to a maximum of 3 items.
Output Format
The final plan should be presented in the exact format specified in the provided template, with no additional prefaces or explanations included in the assistant's message. The plan should stand alone as a clear, actionable guide for the user.
Checklist Item Best Practices
- When naming checklist items, point to specific files or modules where applicable, using syntax like
src/...,app/..., orservices/.... This helps the user pinpoint the exact location of the required changes. - Include concrete validation steps, such as "Run npm test" or "Add unit tests for X". These specific references ground the plan in the user's development environment.
- When relevant, include steps for safe rollout, such as feature flagging, migration plans, or rollback instructions. This demonstrates a comprehensive understanding of the task's implications.
Areas to Avoid
- Steer clear of vague or overly broad checklist items, such as "handle backend" or "do authentication". These terms lack the precision required for an actionable plan.
- Avoid including too many micro-steps that could be combined into a single, more meaningful task.
- Do not insert actual code snippets into the plan; keep the plan implementation-agnostic to allow for the user's preferred coding style and tools.
Your goal is to deliver a plan that is both strategic and tactical, providing the user with a clear roadmap to successfully complete the requested task within their specific codebase.
Original prompt (before our improvements)
--- name: create-plan description: Create a concise plan. Use when a user explicitly asks for a plan related to a coding task. metadata: short-description: Create a plan --- # Create Plan ## Goal Turn a user prompt into a **single, actionable plan** delivered in the final assistant message. ## Minimal workflow Throughout the entire workflow, operate in read-only mode. Do not write or update files. 1. **Scan context quickly** - Read `README.md` and any obvious docs (`docs/`, `CONTRIBUTING.md`, `ARCHITECTURE.md`). - Skim relevant files (the ones most likely touched). - Identify constraints (language, frameworks, CI/test commands, deployment shape). 2. **Ask follow-ups only if blocking** - Ask **at most 1–2 questions**. - Only ask if you cannot responsibly plan without the answer; prefer multiple-choice. - If unsure but not blocked, make a reasonable assumption and proceed. 3. **Create a plan using the template below** - Start with **1 short paragraph** describing the intent and approach. - Clearly call out what is **in scope** and what is **not in scope** in short. - Then provide a **small checklist** of action items (default 6–10 items). - Each checklist item should be a concrete action and, when helpful, mention files/commands. - **Make items atomic and ordered**: discovery → changes → tests → rollout. - **Verb-first**: “Add…”, “Refactor…”, “Verify…”, “Ship…”. - Include at least one item for **tests/validation** and one for **edge cases/risk** when applicable. - If there are unknowns, include a tiny **Open questions** section (max 3). 4. **Do not preface the plan with meta explanations; output only the plan as per template** ## Plan template (follow exactly) ```markdown # Plan <1–3 sentences: what we’re doing, why, and the high-level approach.> ## Scope - In: - Out: ## Action items [ ] <Step 1> [ ] <Step 2> [ ] <Step 3> [ ] <Step 4> [ ] <Step 5> [ ] <Step 6> ## Open questions - <Question 1> - <Question 2> - <Question 3> ``` ## Checklist item guidance Good checklist items: - Point to likely files/modules: src/..., app/..., services/... - Name concrete validation: “Run npm test”, “Add unit tests for X” - Include safe rollout when relevant: feature flag, migration plan, rollback note Avoid: - Vague steps (“handle backend”, “do auth”) - Too many micro-steps - Writing code snippets (keep the plan implementation-agnostic)