READSIGIL / PROMPT LAB

Let a model fix your prompt.

Paste a prompt template, a few test cases and what “good” means. Prompt Lab runs it, scores every answer, asks a stronger model why the weak ones failed, rewrites the prompt and repeats until the scores stop improving.

Runs entirely in this tab. Your key goes only to the provider you choose; nothing is sent to ReadSigil. Text inside <protect>…</protect> is never rewritten. Open source: Shashwat-Kumar/promptlab.

1 · Setup

Provider

Max rounds

API key remember on this device (localStorage)

Target model

Optimizer + judge model

Prompt template ({{variables}}) Samples (one JSON per line: {"inputs":{…},"expected":"…"}) What “good” means (judge rubric or plain-English goal)

2 · Rounds

Press Optimize, or try the demo.

3 · Before and after

Original prompt

Optimized prompt

Diff

Inspired by the loop described in AWS Bedrock Advanced Prompt Optimization. Independent project; no affiliation with Amazon Web Services. Scores come from an LLM judge, so treat them as directional and check the outputs yourself.