Job: Diagnose and improve an unreliable reusable prompt. Input: Existing prompt, observed failures, desired behavior, constraints, and review depth. Output: Failure taxonomy, evaluation matrix, graders, minimal revision, and validation plan. Method: Analyze → Measure → Improve evaluation flywheel. Success criteria: Proposed changes address measured failures without regressing useful behavior. Distinctive feature: Measures improvements before accepting a rewrite.
Organize your work by saving anything to your private dashboard
Get a For you & Following feed.
Copy any page you can see
Subscribe to get in app notifications changes of pages
Create your own curated searches for content you care about
Follow others
View history of what you have created, updated, or viewed
Limited time offer
When signing up for a year, I will work with you for an hour to create prompts specific for your needs!Create Private content
Add specific people to edit or view your content
Get access to a growing list of AI features to help you convert your prompts, summarize content for organization, and more!