RePrompter skill
RePrompter interviews you about a rough request and turns it into a scored, structured prompt or a build prompt for a project.
- 120 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Team modes launch several subagents and burn tokens, and the optional hooks edit `settings.json`.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/reprompterDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Team modes launch several subagents and burn tokens, and the optional hooks edit `settings.json`.
This is third-party code. Review the repository files before installing.
What it does
RePrompter, an agent skill, works in two modes. On `reprompt this` it asks clarifying questions and returns a structured XML prompt with role, context, task, requirements and success criteria plus a quality score. On `one-shot this` it runs a short interview of up to four questions and writes a prompt that builds a product in one session. The result can be pasted, compressed into a `/goal` command, compiled into a runnable Claude `.workflow.js` or sent to another model. A team mode with several agents and a Reverse mode that recovers a prompt from a finished output are included.
Who it is for. Developers and founders who write prompts for Claude Code and Codex in a hurry and want more predictable results.
Good fit when
- A messy request for a dashboard with caching and tests must become a clear brief for an agent.
- A project from scratch is described in one prompt for a full build in a single session.
- A good answer from another model is dissected to recover the prompt that produced it.
- A long task is compressed into a `/goal` command for Codex or Claude Code.
Not a fit when
- You want a normal answer with no prompt rework: the skill triggers only on an explicit request such as 'reprompt this'.
- The task is small and unambiguous: an interview only slows you down.
- You want unattended team runs: spreading work across agents burns many tokens.
Example request
Reprompt this: build a crypto dashboard with CoinGecko data, caching and tests. Ask questions and return a structured prompt with a score.Limitations
The skill triggers only on an explicit request, so an ordinary build request stays ordinary. The `/goal` command in Claude Code needs version 2.1.139 or newer and enabled hooks. The Ambient Prompt Gate hook is off by default, so enable it deliberately. Team modes and autonomous builds start several subagents and spend noticeably more tokens. The author lists rough edges in the maximal build modes himself, and multi-agent results depend on the model and settings.
How to disable. In Claude Code run `/plugin uninstall reprompter@reprompter`. Delete a copied skill from `~/.claude/skills/reprompter`, `~/.codex/skills/` or `~/.hermes/skills/` by hand and remove hooks from `settings.json` if you enabled them.
FAQ
Which phrases trigger it?
`reprompt this` to improve a request and `one-shot this` to build a whole thing. A generic 'build me an app' does not invoke it.
Which agents does it work with?
Claude Code, Codex, OpenClaw, Grok CLI, Hermes Agent or any model that accepts a structured prompt.
What does Reverse mode do?
It takes a good output and reconstructs the prompt structure behind it so you can reuse the technique.
Are hooks on by default?
No. The ambient Prompt Gate is off by default and `/reprompter:reprompter` works without it.
How do I hand the result to another model?
Through headless-relay: the skill packages the prompt and sends it to the chosen model. See the README.
Discussion
No questions or reviews yet.
Related
SkillsA skill that scores a prompt on five criteria, picks one of 31 frameworks and rewrites the request after clarifying questions
A skill that writes precise prompts tailored to a specific tool: Cursor, Claude Code, Midjourney, Stable Diffusion and more
A local prompt coach for Claude Code and Codex: scores a prompt by work stage, rewrites it and suggests one habit to practise.
A hook plugin that adds context to a Claude Code prompt and asks one to six grounded questions when a request is vague.
Claude Code Prompts (RepoWise)
claude-code-prompts
Independently written prompt templates for AI coding agents: system prompts, tool prompts, delegation, memory and multi-agent coordination
A skill that turns a vague request into a Goal Prompt with success criteria and stop conditions, plus a Loop Prompt if needed.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| RePrompter (this card) | no | 120 | ||
| Prompt Optimizer | yes | not required | 41,320 | optimizes a prompt in a single pass; RePrompter interviews first and scores the outcome |
| Autoprompt | no | 1,230 | improves requests automatically without a dialogue; this one adds clarifying questions, a whole-build mode and Reverse |
README in short
Version 13 reduced the work to two intents and folded the old lanes into them, so every earlier trigger routes to one of the two. Templates follow 2026 vendor guidance.