← Back to Skills Marketplace
kevin0818-lxd

IELTS Speaking Coach

by kevin0818-lxd · GitHub ↗ · v1.1.2 · MIT-0
cross-platform ⚠ suspicious
284
Downloads
0
Stars
0
Active Installs
5
Versions
Install in OpenClaw
/install ielts-speaking-coach
Description
IELTS Speaking examiner and tutor. Evaluates spoken English on Fluency, Lexical Resource, Grammar, Pronunciation (Band 1-9). Provides grammar corrections, vo...
README (SKILL.md)

IELTS Speaking Coach

Full-featured IELTS Speaking practice skill with audio pronunciation scoring, mock exams, ZPD learning paths, and adaptive difficulty.

Scope

The skill supports:

  • Part 1 interview practice
  • Part 2 cue-card practice (42 difficulty-tagged cards)
  • Part 3 discussion practice
  • IELTS speaking scoring (text + audio)
  • Grammar correction
  • Vocabulary upgrades
  • Spoken-register model answers
  • ZPD learning path generation (Menu 6)
  • Full mock exam simulation (Menu 7)
  • Adaptive difficulty based on band level

Permissions

Permission Purpose
network LLM API calls for scoring, feedback, and model answer generation
shell ffmpeg audio format conversion for pronunciation scoring from voice messages

Audio Analysis

When the user sends an audio message:

  1. Use ffmpeg to convert to WAV (16kHz mono)
  2. Transcribe using available ASR
  3. Map ASR confidence to PR band score
  4. Include specific pronunciation issues in feedback

If ffmpeg is unavailable, fall back to text-only mode with PR estimated from other criteria.

Entry Mode

Entry phrases:

  • 进入雅思口语模式
  • 启动雅思口语教练
  • 开启雅思口语陪练
  • Use IELTS speaking coach
  • Start IELTS speaking mode

Fixed menu on entry:

  1. Part 1 练习
  2. Part 2 练习
  3. Part 3 练习
  4. 口语评分
  5. 语法纠错
  6. 学习路线
  7. 模拟考试

Practice Flows

Part 1

  • One question at a time, examiner style
  • Topics: study, work, hometown, home, hobbies, daily routine, friends, technology, food, weekends
  • 3 questions default, same topic up to 3 Qs then rotate

Part 2

  • Source priority: cue-cards-2025-may-aug.md → cue-cards.md → generate new
  • Cue card format: 话题 / 你应该说 / 准备1分钟 / 作答1-2分钟
  • After answer, default to scoring template

Part 3

  • Abstract discussion questions, one at a time + follow-up
  • If after Part 2, link to same topic family
  • 3 questions default

Mock Exam (Menu 7)

  • Part 1: 2 topics × 2-3 Qs, no per-answer feedback
  • Part 2: Cue card + follow-ups
  • Part 3: 3-4 abstract Qs linked to Part 2 topic
  • Final report: 总分, 分Part评分, 单项分, 强项, 薄弱环节, ZPD学习方向, 考试建议

Learning Path (Menu 6)

  • Ask for transcript or band scores
  • Output: 当前定位, ZPD词汇目标, ZPD语法目标, 话题词块, 每日建议, 阶段目标+周期
  • Reference: learning-path.md

Feedback Templates

Scoring Template (Menu 4)

  1. 总分
  2. 单项分 (FC/LR/GRA/PR, mark PR source: audio or estimated)
  3. 评分依据
  4. 主要问题 (≤3)
  5. 提升建议 (≤3)
  6. 参考改写
  7. 下一步学习方向 (ZPD recommendations)

Grammar Correction Template (Menu 5)

  1. 原句
  2. 修改后
  3. 错误说明
  4. 更自然的口语表达
  5. 一句话建议

Scoring Policy

  • Score FC, LR, GRA, PR independently with evidence
  • Band 1-9, 0.5 increments
  • CHAI calibration before averaging
  • Audio input: PR from ASR confidence mapping
  • Text-only: PR = median(FC, LR, GRA), note "estimated"
  • Band descriptors: scoring-rubric.md

Adaptive Difficulty

Band Question Style
4-5 Familiar daily topics, concrete cue cards, simple Qs
5.5-6.5 Less common topics, moderately abstract cards, comparison Qs
7+ Nuanced topics, abstract cards, hypothetical/policy Qs

Supporting Files

  • scoring-rubric.md — IELTS band descriptors
  • cue-cards-2025-may-aug.md — 15 official 2025 May-Aug cue cards
  • cue-cards.md — 42 difficulty-tagged cue cards (7 categories)
  • learning-path.md — ZPD vocabulary/grammar progression
  • pronunciation-guide.md — Chinese speaker pronunciation guide
  • vocab-map.json — Topic-aware vocabulary upgrades
  • examples.md — Sample interactions

Self-hosted Backend (Optional, GitHub only)

The GitHub repository includes an optional backend/ directory with a FastAPI server providing enhanced features (DL-based scoring, persistent learning state, vocabulary ontology). The backend is not included in the ClawHub package and is not required — all core features work via the built-in LLM and bundled reference files.

Usage Guidance
This skill appears coherent for an IELTS speaking tutor. Before installing, consider: (1) audio privacy — the skill converts and transcribes voice messages (ffmpeg + ASR); confirm whether your platform's ASR/LLM will keep audio local or send it to external services and whether you consent to that; (2) shell permission scope — the SKILL.md says shell is only for ffmpeg, but granting shell access can be powerful on some platforms, so verify the platform enforces that restriction; (3) optional backend — a self-hosted backend exists on GitHub but is not included in the ClawHub package; only enable or deploy that backend if you review its code and trust the host; (4) content/copyright — bundled cue-card files include recent exam bank items; ensure you are comfortable with their use. If you need higher assurance, ask the maintainer for an explicit statement about where audio/ASR processing happens (local vs remote) and any network endpoints used for ASR or telemetry.
Capability Analysis
Type: OpenClaw Skill Name: ielts-speaking-coach Version: 1.1.2 The skill requests 'shell' and 'network' permissions to perform audio processing and LLM scoring. Specifically, 'skill.yaml' and 'SKILL.md' instruct the agent to use the shell to execute 'ffmpeg' for converting user audio messages. While this capability is aligned with the stated purpose of an IELTS coach, granting an AI agent shell access introduces a high-risk attack surface for command injection. The 'CHANGELOG.md' shows the author has actively worked to resolve previous safety flags by removing external backend code from the bundle, but the inherent risk of the shell permission warrants a suspicious classification.
Capability Assessment
✓ Purpose & Capability
Name/description (IELTS speaking coach) match the packaged files and SKILL.md: scoring rubrics, cue cards, learning-path, pronunciation guide and examples. The declared uses of network (LLM) and shell (ffmpeg audio conversion) are appropriate for the stated features (LLM-generated feedback and optional audio scoring). The optional GitHub backend is documented as separate and not included in the published package.
ℹ Instruction Scope
SKILL.md instructs the agent to convert user audio with ffmpeg and to transcribe using an available ASR; scoring maps ASR confidence to a pronunciation band. That behavior is consistent with the feature set, but it implies user audio may be processed/transmitted to an ASR or LLM endpoint. The instructions do not request reading unrelated files or environment variables. Users should note that audio data may leave the device depending on platform ASR/LLM configuration.
✓ Install Mechanism
No install spec or runtime downloads are present in the ClawHub package. The package is instruction-only with bundled reference files. The optional backend is explicitly excluded from the published package and described as self-hosted on GitHub only.
✓ Credentials
The skill requires no environment variables, no external credentials, and no config paths. Declared permissions (network for LLM calls, shell for ffmpeg) are proportionate to audio scoring and LLM-based feedback. There are no unexplained SECRET/TOKEN requests.
✓ Persistence & Privilege
always is false and the skill is user-invocable. The skill does not request permanent presence or system-wide configuration changes. The documented optional backend is separate and not installed by the skill.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install ielts-speaking-coach
  3. After installation, invoke the skill by name or use /ielts-speaking-coach
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.1.2
Safety fix: exclude backend code from ClawHub package to resolve all safety flags. Shell permission is for ffmpeg audio conversion only.
v1.1.1
Safety fix: exclude backend code from ClawHub package, clarify shell permission is for ffmpeg only.
v1.1.0
Add audio pronunciation scoring via ffmpeg+ASR, mock exam simulation, ZPD learning paths, adaptive difficulty, pronunciation guide, expanded 42 cue cards.
v1.0.1
Add fixed-entry mode, menu-driven IELTS practice flows, and standardized scoring/correction templates.
v1.0.0
Initial release: four-criterion scoring, CHAI calibration, vocab upgrades, model answers
Metadata
Slug ielts-speaking-coach
Version 1.1.2
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 5
Frequently Asked Questions

What is IELTS Speaking Coach?

IELTS Speaking examiner and tutor. Evaluates spoken English on Fluency, Lexical Resource, Grammar, Pronunciation (Band 1-9). Provides grammar corrections, vo... It is an AI Agent Skill for Claude Code / OpenClaw, with 284 downloads so far.

How do I install IELTS Speaking Coach?

Run "/install ielts-speaking-coach" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is IELTS Speaking Coach free?

Yes, IELTS Speaking Coach is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does IELTS Speaking Coach support?

IELTS Speaking Coach is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created IELTS Speaking Coach?

It is built and maintained by kevin0818-lxd (@kevin0818-lxd); the current version is v1.1.2.

More Skills
self-improving agent
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
⬇ 463815 · by pskoett
Skill Vetter
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope, and suspicious patterns.
⬇ 259802 · by spclaudehome
Polymarket
Query Polymarket prediction markets. Check odds, find trending markets, search events, track price movements.
⬇ 232503 · by joelchance
Self-Improving + Proactive Agent
Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
⬇ 200680 · by ivangdavila
Github
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
⬇ 191345 · by steipete
ontology
Typed knowledge graph for structured agent memory and composable skills. Use when creating/querying entities (Person, Project, Task, Event, Document), linkin...
⬇ 190333 · by oswalpalash

💬 Comments