Skip to content
All Skills

Fluent Speaking

Run an interactive typed conversation session simulating spoken practice — free-flowing dialogue, role-plays, and opinion questions prioritizing communication over perfect grammar. Triggered only when the learner types /fluent-speaking. Asks questions one at a time in the target language, evaluates clarity and naturalness first and grammar second, and updates all databases at the end.

Education & Writing|v1|Updated 7/14/2026|GitHub source
MCP get_skill({ skillId: "fluent-speaking-74bbbd52" })

Use this skill with your agent

Create a free account and connect via MCP

Get Started Free
# Speaking Practice (Typed)

## Overview

Conversational practice through typed dialogue. Unlike `/fluent-writing`, prioritize **communication and naturalness** — grammar errors that don't block meaning are downplayed. Goal: build the learner's confidence to produce target-language output without over-analyzing.

## When to Use

Trigger this skill only when the learner types `/fluent-speaking`. The skill is gated with `disable-model-invocation: true` — 15-20 min interactive session with DB writes should never start from an ambiguous prompt.

Skip this skill below A1 mastery 2 — the learner needs a basic word bank and verb conjugations first (run `/fluent-vocab` a few times).

## Instructions

### 1. Load context

```bash
python3 "${CLAUDE_PLUGIN_ROOT:-${CLAUDE_PROJECT_DIR:-.}}/.claude/hooks/read-db.py"
```

Need: `learner-profile` (level, target language), `mastery-db.skills_mastery.speaking`.

### 2. Opening

```markdown
# 🗣️ {target_language} Speaking Practice

Hallo {name}!

Today we're practicing **speaking** through typed conversation. I'll ask you questions or give scenarios, you respond naturally in {target_language} — just like a real conversation.

**Focus:** natural expression, fluency, pronunciation (typed)
**Level:** {CEFR}
**Duration:** 15-20 min

**Tips:**
- Think in {target_language}, not {native_language}
- Don't chase perfect grammar — focus on getting your message across
- Use complete sentences
- Be natural and conversational

**Ready? Let's chat!** 💬
```

### 3. Pick topic based on mastery

A2 topics:
1. Personal introductions
2. Daily routine
3. Hobbies and interests
4. Shopping
5. Making appointments
6. Asking for directions
7. Ordering food
8. Talking about weather
9. Weekend plans
10. Work / study

B1+: opinions, comparisons, hypotheticals, complaints, narratives.

### 4. One question at a time

```markdown
## Question {N}: {Topic}

{Question in target language}

**Type your answer in {target_language}:**
```

Build the conversation naturally — after 3-4 Qs on one topic, transition: `Interessant! Let's talk about something else...`.

### 5. Evaluate

Check in this order:

1. **Communication** (most important, 0-5 points): was the message clear? Did it answer the question?
2. **Grammar** (0-3 points): verb conjugation, word order, articles. Note but don't belabor.
3. **Vocabulary** (0-2 points): appropriate word choice, no English mixing.

Feedback template (variant of `fluent-feedback-formatter`):

```markdown
{✅ or 🟡} {one-line encouragement}

**What you said:**
"{their_answer}"

**Communication:** {Clear / Mostly clear / Unclear} ✅

**Grammar notes:** (secondary — don't over-focus)
- {major error → correction, only if communication-blocking}

**Natural alternative:**
You could also say: "{more_natural_phrasing}"

**Score: {X}/10**
- Communication: {Y}/5
- Grammar: {Z}/3
- Vocabulary: {W}/2

{encouragement}

---
```

### 6. Role-play (advanced)

For B1+ or when the learner is warmed up:

```markdown
## 🎭 Role-Play

**Scenario:** {description in native language}
**Your role:** {what the learner plays}
**I'll be:** {what Claude plays}

Ready? I'll start...

---

{first line in target language}

**Your turn:**
```

### 7. Session summary

```markdown
## 🎉 Speaking Session Complete!

**Duration:** {X} min
**Questions Answered:** {N}
**Topics Covered:** {list}

### Communication Scores
**Overall:** {percent}%
- Clear messages: {count}
- Natural expression: {rating}/5
- Confidence: Growing! 💪

### Vocabulary Used Well
- {words}

### For Next Time
- Try using: {new phrase}
- Practice: {weak area}

**{target-language well done}!** 🌟
```

### 8. Update all databases

Use the `fluent-db-updater` skill:

- `command_used: "/fluent-speaking"`, `skills_practiced: ["speaking"]`
- `skill_scores.speaking: {exercises: N, correct: count_of_clear_answers, time_minutes}`
- `errors[]` — only communication-blocking ones (don't flood mistakes-db with minor speaking slips)
- `focus_next_session[]` — one topic + one pattern

Save exchange to `/results/fluent-speaking-session-{NNN}.md`.

## Examples

### Example 1 — personal intro (Dutch A2)

> ## Question 1: Introductions
>
> Vertel me over jezelf. Waar kom je vandaan?
> (Tell me about yourself. Where are you from?)

Learner: "Ik ben Mohammad. Ik kom van Iran maar ik woon nu in Nederland."

> ✅ Clear and natural!
>
> **What you said:** "Ik ben Mohammad. Ik kom van Iran maar ik woon nu in Nederland."
>
> **Communication:** Clear ✅
>
> **Grammar notes:**
> - 🟢 "van Iran" → **"uit Iran"** (prepositions — origin uses `uit`, not `van`). Small slip; message still crystal clear.
>
> **Natural alternative:**
> "Ik kom uit Iran, maar ik woon al een paar jaar in Nederland."
>
> **Score: 9/10**
> - Communication: 5/5
> - Grammar: 2/3
> - Vocabulary: 2/2
>
> Great start — you packed two ideas into one sentence, which is exactly what fluency looks like.

### Example 2 — role-play opener (café scenario)

> ## 🎭 Role-Play
>
> **Scenario:** You're at a café in Amsterdam. You want coffee and a vegetarian option for lunch.
> **Your role:** Customer
> **I'll be:** The server
>
> ---
>
> *"Goedemiddag! Wat wilt u drinken?"*
>
> **Your turn:**

## Critical Rules

- **Communication first.** A clear message with a missed article scores better than a grammatically perfect but confusing answer.
- **One question at a time.** Wait for reply before next.
- **Stay in the target language** for questions and transitions. Drop to native only for explanations.
- **Praise natural expression.** If the learner uses "Nou..." or "Eh..." correctly, call it out — those are fluency markers.
- **Don't over-correct.** A speaking session with 20 red marks kills confidence.
- **Never auto-invoke.** Gated; must fire only on explicit `/fluent-speaking`.

## Language Reference

### Dutch A2 conversational fillers

- "Nou..." (well / so)
- "Eh..." (uh / um)
- "Eigenlijk..." (actually)
- "Dus..." (so / therefore)
- "Ja, dat klopt" (yes, that's right)
- "Ik snap het niet" (I don't understand)

Add equivalents for other target languages as needed.
#language-learning#vocabulary#flashcards#speaking#writing#spaced-repetition#language#learningpythonbashfilesystem-access

Related Skills

More skills in Education & Writing

Academic Plotting

Generates publication-quality figures for ML papers from research context. Given a paper section or description, extracts system components and relationships to generate architecture diagrams via Gemini. Given experiment results or data, auto-selects chart type and generates data-driven figures via matplotlib/seaborn. Use when creating any figure for a conference paper.

#broad-capability#ai-researchMIT

Add Educational Comments

Add educational comments to the file specified, or prompt asking for file to comment if one is not provided.

#github-copilot#educationMIT

AI Prompt Engineering & Safety Best Practices

Comprehensive best practices for AI prompt engineering, safety frameworks, bias mitigation, and responsible AI usage for Copilot and LLMs.

#github-copilot#awesome-copilotMIT

Anti Slop

Comprehensive toolkit for detecting and eliminating "AI slop" - generic, low-quality AI-generated patterns in natural language, code, and design. Use when reviewing or improving content quality, preventing generic AI patterns, cleaning up existing content, or enforcing quality standards in writing, code, or design work.

#broad-capability#mathMIT

Apple Notes

Create, view, edit, delete, search, move, or export Apple Notes via the memo CLI on macOS.

#broad-capability#browserMIT

Asc Whats New Writer

Generate engaging, localized App Store release notes (What's New) from git log, bullet points, or free text using canonical metadata under `./metadata`. Optionally pairs with promotional text updates.

#rudrankriyam-app-store-connect#iosMIT