Skip to content
Skills Library
AI & agents · Writing

reverse-prompt

/reverse-prompt

Use when the user wants a single conversational prompt that would rebuild the current project from scratch. Three modes — Quick (file tree + names), Deep (full understanding), Focus (one feature only).

Repository
sift-coder ↗
Licence
MIT
Last repository change
03.06.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 03.06.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Use when the user wants a single conversational prompt that would rebuild the current project from scratch. Three modes — Quick (file tree + names), Deep (full understanding), Focus (one feature only).

Original author description

Use when the user wants a single conversational prompt that would rebuild the current project from scratch. Three modes — Quick (file tree + names), Deep (full understanding), Focus (one feature only).

How it works

  1. Modes
  2. Process
  3. Output shape (Deep)
  4. Output shape (Quick)
  5. Output shape (Focus)
  6. Anti-patterns
Original SKILL.md headings · 69 lines, 322 words
SKILL.md preview
---
name: reverse-prompt
description: Use when the user wants a single conversational prompt that would rebuild the current project from scratch. Three modes — Quick (file tree + names), Deep (full understanding), Focus (one feature only).
---

# Reverse-prompt skill

Generate a single conversational prompt — what you'd say to an agent to rebuild this project. Useful for: handoffs, "starting over from scratch", spec extraction, onboarding new contributors.

## Modes

- **Quick** — just structure (file tree, top-level names). Cheapest. ~500 tokens.
- **Deep** — full understanding (architecture, key files, data flow, conventions). ~3-5k tokens.
- **Focus <area>** — one feature/area only, with surrounding context. ~1-2k tokens.

## Process

1. **Determine mode.** If user says "quick", "deep", "focus on X" — use that. Otherwise default Deep.
2. **Cache check.** Compute fingerprint = sha1(top-level tree + manifest hash). Check `~/.siftcoder/reverse-prompt-cache/<fingerprint>.json`. If &lt; 24h old and matching mode — use it.
3. **Scan.**
   - Quick: `ls`, `cat package.json`, `cat README.md`. That's it.
   - Deep: launch Explore subagent to map architecture; read 5-10 key files; sample test files; check git log.
   - Focus: launch Explore on the named area only; ignore everything else.
4. **Synthesise** the prompt as second-person ("Build me a project that..."). Include:
   - Goal (1 sentence)
   - Stack (lang/framework/runtime)
   - Folder structure (compressed)
   - Key abstractions (named, with one-line responsibility each)
   - Conventions (naming, error handling, test style)
   - Build/test commands
   - Constraints / non-goals
5. **Cache** the result with fingerprint + ts.

## Output shape (Deep)

```
Build me a <one-line description>.

Stack: <lang> <framework> <runtime>

Project structure:
<compressed tree>

Key modules:
- <name>/ — <responsibility>
…

Conventions:
- <pattern 1>
- <pattern 2>

Build:    <cmd>
Test:     <cmd>
Lint:     <cmd>

Non-goals:
- <thing it deliberately doesn't do>
```

## Output shape (Quick)

A short paragraph + bare structure listing.

## Output shape (Focus)

Like Deep but scoped to the area, plus a "wires-into:" line listing what surrounds it in the parent project.

## Anti-patterns

- Don't just `cat` files into the prompt. Synthesise.
- Don't include secrets, env vars, or local paths.
- Don't generate &gt; 5k tokens for Deep — if the project is big, summarise harder.

Versions

The database has no other skills named reverse-prompt.

Database snapshot: 13.11.2026 · package version b12026fe7c50 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100reverse-prompt-engineer
    Reconstruct one plausible, evidence-led, copy-ready user-level prompt from an existing AI output and its context: a report, article, analysis, code, image, video, audio, website or design, or a long conversation with files. Use whenever the user asks what prompt produced,…
  • 100prompt-craft
    USE WHEN improve prompt, write prompt, craft prompt, analyze prompt, prompt engineering, subagent prompts. Analyze, craft, and improve prompts using 19 research-backed techniques. Modes - Analyze (critique existing), Craft (build from scratch), Teach (explain techniques), Quick…
  • 100reprompter-reverse
    RePrompter Reverse mode (advanced) — extract the prompt DNA from a great output and emit the XML prompt that would have produced it, with extracted success criteria; optionally seeds the Prompt Flywheel. Use when: "reverse reprompt", "reverse reprompter", "prompt dna", "prompt…
  • 87json-prompt-engine
    Analyze a reference image (or video frame) and emit a ready-to-paste structured JSON prompt that captures every visual quality needed to reproduce it through `ralphy generate image --prompt-file`. The reverse of the prompt cookbook: image in -> dense JSON prompt out (scene /…
  • 100prompt
    Build a great prompt through a conversational interview. Use this skill whenever the user wants help writing, improving, or crafting a prompt — including phrases like "help me write a prompt", "I need a prompt for", "how do I prompt Claude to", "make me a prompt that", "I want…
  • 95Research
    Comprehensive research and content extraction — quick/standard/extensive/deep modes with multi-agent parallel research, content retrieval, AI trends analysis, and 242+ Fabric patterns. USE WHEN research, do research, quick research, extensive research, deep investigation, find…

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.