Hook · Prompt Playbook Studio
A clearer prompt before you start
Adds a brief context note when you're drafting a prompt that seems to lack a clear target or a way to check success.
When it steps in
When you send a request in Codex or Claude Code, it checks whether you are asking for help writing a prompt.
Codex + Claude Code · Enable in your AI app
The plugin and its hooks must be enabled and trusted in your AI app. Ordinary requests pass through without a hint.
What you can expect
The assistant receives a short reminder to clarify the target or success check when one appears to be missing. Your submitted words stay unchanged.
How the workflow unfolds
- 01
Read the prompt
Takes your latest submitted prompt text, capped at 8,000 characters.
- 02
Check for gaps
Uses local pattern matching to see if the prompt is about prompt-writing and is missing a target or a proportionate success check.
- 03
Add a hint if needed
If a gap is found, adds a short context note to the host; otherwise stays silent.
You stay in control
Use your AI app’s hook or plugin settings to control whether this runs. The script has no separate per-project switch.
Know the limits
This is a small pattern check, not a full prompt review. It reads at most 8,000 prompt characters and may miss intent. It does not save your prompt, rewrite it, call other tools, or block the turn.
Technical details and sources
Based on version 0.3.1. These declarations describe the package; they do not confirm that your AI app has enabled or run the hook.
Show events and release source files
- Codex
- UserPromptSubmit
- Claude Code
- UserPromptSubmit