F-xx is a structured functional requirement id (e.g., F-01) with fields like said (verbatim user words), verify (machine-checkable), and priority.

Where is the output stored?

The Prompt JSON is saved to .copilot-tracking/prompt-{NNN}.json.

What happens if the user input is ambiguous?

Ambiguities trigger a mandatory Clarification step; gaps are not filled with assumptions and may be marked as TBD for later confirmation.

prompt

Scanned

npx machina-cli add skill Roberdan/MyConvergio/prompt --openclaw

Files (1)

SKILL.md

3.0 KB

Prompt Translator Skill

Reusable workflow extracted from prompt agent expertise.

Purpose

Extract structured functional requirements (F-xx) from natural language user input, producing machine-readable JSON for downstream planning agents.

When to Use

New feature requests that need structured decomposition
Ambiguous user input requiring clarification before planning
Requirements extraction before @planner execution
Any input that needs F-xx requirement mapping

Workflow Steps

Context Gathering
- Read repository state via git-digest.sh
- Understand project structure and existing patterns
Clarification (MANDATORY)
- Identify ambiguities in user input
- Ask about: scope, negative requirements, edge cases, priority
- NEVER fill gaps with assumptions — ask or mark TBD
Requirement Extraction
- Extract EVERY requirement (explicit + implicit) as F-xx
- Use EXACT user words — NEVER paraphrase
- Each requirement needs: id, said (verbatim), verify (machine-checkable), priority
Output Generation
- Save to .copilot-tracking/prompt-{NNN}.json
- Include: objective, user_request, requirements[], scope, stop_conditions
User Confirmation
- Ask "Have I captured everything? Anything missing?"
- User confirms → offer handoff to @planner

Inputs Required

User request: Natural language description of desired feature/change
Repository context: Auto-detected from current working directory

Outputs Produced

Prompt JSON: .copilot-tracking/prompt-{NNN}.json with structured F-xx requirements
Handoff: Ready for @planner to create execution plan

Output Format

{
  "objective": "one sentence goal",
  "user_request": "EXACT user words, verbatim",
  "requirements": [
    { "id": "F-01", "said": "exact words", "verify": "how to check", "priority": "P1" }
  ],
  "scope": { "in": ["included"], "out": ["explicitly excluded by user"] },
  "stop_conditions": ["All F-xx verified", "Build passes", "User confirms"]
}

Critical Rules

Rule	Requirement
said	EXACT user words, never paraphrase
verify	Machine-checkable (grep, test command, build passes), not prose
scope.out	ONLY items USER explicitly said to exclude, NEVER add on own initiative
Purpose	This JSON is read by planner agent to generate spec.json

Related Agents

po-prompt-optimizer - Prompt engineering and optimization
strategic-planner - Strategic planning from requirements
ali-chief-of-staff - Orchestration and delegation

Source

git clone https://github.com/Roberdan/MyConvergio/blob/master/.claude/skills/prompt/SKILL.mdView on GitHub

Overview

The Prompt Translator Skill extracts structured functional requirements (F-xx) from natural language user input and outputs a machine-readable JSON for downstream planning agents. It preserves the exact user words, enforces mandatory clarification, and saves results to .copilot-tracking as prompt-{NNN}.json to hand off to @planner.

How This Skill Works

It gathers repository context (e.g., via git-digest.sh), identifies ambiguities, and prompts for mandatory clarifications. Then it extracts every requirement (explicit + implicit) as F-xx, with fields id, said, verify, and priority, using the exact user words. Finally, it writes the Prompt JSON to .copilot-tracking/prompt-{NNN}.json and asks the user to confirm before handing off to the planner.

When to Use It

New feature requests that require structured decomposition into F-xx requirements.
Ambiguous user input requiring clarification before planning.
Before executing @planner to ensure requirements are well-defined.
Any input that needs precise F-xx requirement mapping.
When a machine-readable spec is needed for downstream automation.

Quick Start

Step 1: Read the user request and repository context (e.g., via git-digest.sh).
Step 2: Enforce mandatory clarification and extract every F-xx requirement using exact user words.
Step 3: Save the Prompt JSON to .copilot-tracking/prompt-{NNN}.json and prompt the user for confirmation.

Best Practices

Preserve exact user words; do not paraphrase the input.
Proactively request scope, edge cases, and priority in clarifications.
Define verify as machine-checkable commands or tests, not prose.
Mark unresolved gaps as TBD rather than guessing.
Save outputs to .copilot-tracking/prompt-{NNN}.json and obtain user confirmation.

Example Use Cases

User says: 'Add an OAuth login flow with Google and GitHub' -> F-01, said: 'Add an OAuth login flow with Google and GitHub', verify: 'grep for auth routes and tests', priority: P1.
User says: 'Support three new locales' -> F-02, said: 'Support three new locales', verify: 'update i18n files and locale tests', priority: P2.
User says: 'Improve app startup time' -> F-03, said: 'Improve app startup time', verify: 'benchmark startup duration before/after', priority: P1.
User says: 'Clarify whether admin privileges are required for the feature' -> F-04, said: 'Clarify admin requirement', verify: 'confirm with product', priority: P3.
User says: 'Save the plan to .copilot-tracking/prompt-003.json' -> F-05, said: 'Save the plan to .copilot-tracking/prompt-003.json', verify: 'file exists at path', priority: P2.

Frequently Asked Questions

Add this skill to your agents