My Plan Template

I’ve been using this prompt to get Cursor to output a nicely formatted plan

Add to your rules for ai or .cursorrules and ask Cursor to make a plan

Plan Template

# Summary

# Assumptions

# References

- Files `dir/file`
- Libraries `http://library.com`
- Documentation `http://docs.com`

# Implementation Steps

1. First step (`dir/file`)

# Confidence Level

- **High/Medium/Low** confidence in X

# Questions
2 Likes

Can you share an example of how it helped? Also, can cursor read internal documents or they have to publicly accessible?

In the references it will tell me what files it’s looking at and if I’m adding a new dependancy a link to the docs so I can review.

Screenshots of what it looks like in action.

I generally iterate on the plan until the confidence levels increase and I’ve answered all the questions.

I’m using this with my plan vs act mode Plan vs Act modes


i constantly have to do this as part of my workflow even tho i have project task list

you’ve inspired me to create a new tool in my .cursorrules file

<ProposeSolution>
  1. Dont write any code.
  2. Clearly Define the Problem/Feature.
  3. List specific, verifiable assumptions.
  4. List acceptance criteria
  5. List affected files and sub files using tools to identify indirect impacts.
  6. Does this create or reduce technical debt?
  7. Propose a range of potential solutions
  8. Solution Comparison:
     - Create a table comparing solutions based on:
       - Pros and cons
       - Adherence to KISS, DRY, YAGNI.
       - Performance implications
       - Scalability concerns.
       - Maintainability and readability.
       - Security considerations.
       - Development effort.
       - Assign initial confidence score to each solution
  9. Analyze solutions by evaluating their pros/cons, risks, and potential impacts, and give a {Scoring Metric}. Include in the analysis "What could go wrong?".
  10. Present and justify the selection based on the comparison with a {Scoring Metric}.
  11. If the solution is not clear, ask for more information.

  Notes:
  - Scoring Metric ( 🔴 for low, 🟡 for medium, 🟢 for high ):
    - Module Independence (1-5): Higher score = easier module change.
    - Clarity of Code (1-5): Higher score = code is easy to understand.
    - Component Reusability (1-5): Higher score = code is easily reused.
    - Test Coverage (1-5): Higher score = more code is tested.
  - Consider Visual Aids by adding diagrams (UML, flowcharts) to illustrate complex solutions.
</ProposeSolution>

and i can call that tool with the command:

 /ps: propose solution using <ProposeSolution> tool