Editable Prompt Templates with Placeholders

Feature Requests
1

Feature Request: Editable Prompt Templates with Placeholders

Summary:
Support project-based prompt templates (committed in the repo) with editable placeholders, usable directly in Cursor’s agent or chat.

Problem:
Engineers often reuse prompts for common tasks (e.g., writing tests, debugging). cursor.rules help but can’t be easily edited or customized per use in the chat.

Proposed Solution:

  • Allow storing prompt templates in the repo (like cursor.prompts.json).
  • Support placeholders like <ComponentName> or <BugDescription> that users can quickly fill.
  • Enable in-chat editing without altering the original template.

Benefits:

  • Boosts productivity and consistency.
  • Makes prompting easier for new joiners.
  • Encourages sharing effective workflows.

Tip: you can make manual rules be inline editable, and support placeholders.

2

This seems like a need for an extension.

3

I achieved something similar with

Prompt the user for the following information one bit at a time: The following lines are for the AI to use and are in the form "Name, user prompt => replacement token", or " userprompt => replacement token"

* Entity, Database Entity => [entity]
* Schema table, Table => [table]
* Is the entity audited => [entity_audited]
* Domain Object, What is the domain object this represents => [domainobject]
* Module, What module should this go in? => [module]
* Top Level, What top level grouping inside the module should it go => [top_level]
* Are [domainobject] searchable => [searchable]
* Is [domainobject] scoped to an organisation => [org_scoped]

And it does what you expect.