Claude Language Coach — AI skill for Claude Code
Writing feedback on every Claude Code prompt, in the language you are practising, reviewed out of band by a small model.
How to install Claude Language Coach
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open sergiojrdotnet/claude-language-coach and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Claude Language Coach does
Writing feedback on every Claude Code prompt, in the language you are practising, reviewed out of band by a small model.
Alternatives in AI
- Anthropic SDK TypeScript — Access to Anthropic's safety-first language model APIs in TypeScript 2.1k ★
- Cwc Long Running Agents — Claude Code's built-in /goal command gives you a generator/evaluator loop out of the box: set a completion con 668 ★
- Anthropic SDK Csharp — Access to Anthropic's safety-first language model APIs in C# 317 ★
README
Language Coach for Claude Code
Get writing feedback on every prompt you type, in the language you are practising. A small model reviews each prompt in a separate call and shows the fixes under it. Nothing is added to the main conversation, so its context stays clean. Each review is a small, separate request that counts against your usage (see [Cost and privacy](#cost-and-privacy)).
https://github.com/user-attachments/assets/4a21de20-32ec-4cdb-b7a4-b9ff78d2b93b

Each fix also comes with a short explanation of the rule behind it:
- Fullscreen terminal or desktop app: hover over the fix to see the explanation in a card.
- Other surfaces: the explanation prints as a dim
↳line under the fix.
To always print it inline, or to hide it, change the **Explanations** setting (see [Settings](#settings)).

Fixes stay attached to their prompts after `--resume`.
Install
In a Claude Code terminal session:
/plugin install language-coach --marketplace sergiojrdotnet/claude-language-coach
Answer `y` to add the marketplace, then pick the **user** scope so the coach runs in every session.
Requires Claude Code **2.1.291 or newer** (mods / function hooks). The mods API is early access and can change between releases.
Use
| You do | You get |
|---|---|
| Type a prompt | About 1–2 s later, any fixes appear under your prompt. A clean prompt shows nothing. |
/language-coach |
A pane with your history: corrections by category, a 30-day trend of fixes per prompt, your recurring mistakes, and recent fixes. It has Pause and Clear history buttons. |
/language-coach off / /language-coach on |
Pause or resume the coach for the current sess |
Related Skills
Usage Band Mod
Claude Code mod: a live usage band above the prompt with context window, 5-hour, weekly and per-model limits a
Usage Band
A live status band above the Claude Code prompt: model, context, plan limits, real prompt-cache status and cos
Quackd
🦆🧠 Give your Microduck a brain. Tell a small robot with two legs what you want in plain language. An LLM (Cl
Agentic KVM
Give AI agents (Claude Code/Cowork, ChatGPT/Codex, Gemini/Antigravity) bare-metal, out-of-band control of any
A2z Ghost Kernel
Out-of-Band Shadow Sentinel, Mathematical Reflex Governor & Black Box Flight Recorder for Autonomous AI Agents
Apronagents
Multi-agent AI coding with a human merge gate — sandboxed git for agents, small reviewed diffs, your call on e
Related Agents
Mid Task
Executes moderately complex tasks (focused multi-file edits, contained bug fixes, small features, test writing
Tutor Sv
Swedish-language coach for the citizenship requirements — reading and listening at B1, writing and speaking at
Poteto Fable
Third-family writing delegate. Exists so arena and architect can fan out across genuinely different models ins