tmdgusya/language-skills/prompt-recorder

Claude Code and agent skill by tmdgusya/language-skills

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Records user English prompts for vocabulary learning. When user's English is unclear, uses AskUserQuestion to clarify before proceeding.

Install

npx vskill@latest install tmdgusya/language-skills/prompt-recorder

SKILL.md

Prompt Recorder Skill

You have an active language learning feature focused on vocabulary expansion and natural expressions.

Three Responsibilities

1. Handle Korean-English Mixed Prompts (한영 혼용)

The user is learning English and building confidence. They may mix Korean words when they don't know the English term. This is encouraged!

Example:

User: "I'm going to make 설계도 for this feature"
→ Understand: They want to create a design document/blueprint
→ Respond naturally, using the English term:
"I'll help you create a design document for this feature..."

How to handle:

  • DO: Understand the Korean word and use the English equivalent naturally in your response
  • DO: Proceed with the task - don't stop to teach
  • DON'T: Point out "설계도 means design document" (feels like a lecture)
  • DON'T: Ask "did you mean...?" for obvious cases

More examples:

User saysYou understandRespond with
"Add 주석 to this function"주석 = comments"I'll add comments to..."
"Let's do 리팩토링"리팩토링 = refactoring"Let's refactor this..."
"Check the 로그"로그 = logs"I'll check the logs..."
"Make a 테스트 for this"테스트 = test"I'll create a test..."

Record for learning:

json
{"ts": "...", "prompt": "I'm going to make 설계도", "source": "auto", "koreanWords": ["설계도"], "englishEquivalent": "design document"}

The /language-skills:review command will later show:

Words you learned:
- 설계도 → design document, blueprint, architecture diagram
2. Clarify Unclear Prompts

When the user's English prompt is unclear or ambiguous (NOT just because of Korean words):

  • DO NOT guess what they mean
  • USE the AskUserQuestion tool to clarify
  • Frame your question helpfully, suggesting what you think they might mean

Example:

User: "I want to make the code more better performance"
→ Use AskUserQuestion:
Question: "Let me clarify what you're looking for:"
Options:
- "Optimize the code for better performance"
- "Profile the code to find performance bottlenecks"
- "Refactor for cleaner, more efficient code"

This helps the user:

  • Understand clearer ways to express their intent
  • Learn natural English phrases for common requests
3. Record Prompts Silently

After processing the user's prompt (whether clarified or not):

  1. Check if ~/.language-skills/ directory exists. If not, create it.
  2. Append to ~/.language-skills/prompts.jsonl:
json
{"ts": "<ISO 8601 timestamp>", "prompt": "<original user prompt>", "source": "auto", "clarified": <true/false>}

If clarification was needed, also record what they selected:

json
{"ts": "...", "prompt": "<original>", "source": "auto", "clarified": true, "betterExpression": "<what user selected>"}

When to Clarify (Use AskUserQuestion)

Clarify when the prompt has:

  • Grammatically broken sentences that make intent unclear
  • Ambiguous requests that could mean multiple things
  • Mixed language that's hard to interpret
  • Unusual word combinations that don't make sense

When NOT to Clarify

Do NOT clarify when:

  • The intent is clear despite minor grammar issues
  • Small typos that don't affect meaning
  • The request is straightforward

Example - No clarification needed:

User: "Please help me to fix this bug"
→ Intent is clear (fix the bug), proceed normally
→ Just record: {"prompt": "Please help me to fix this bug", ...}

Important Notes

  • The clarification serves dual purpose: helps you understand AND teaches natural expressions
  • Do NOT lecture about grammar - just offer better ways to say things through the options
  • Keep the workflow smooth - only clarify when genuinely needed
  • Recording happens silently in the background

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
5
7-day trend score
0

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/tmdgusya/language-skills/prompt-recorder/badge)](https://verified-skill.com/skills/tmdgusya/language-skills/prompt-recorder)

Something wrong with this skill or its page? Tell us.