chuckwu0

Larkway — Development skill for Claude Code

Development community

Turn your Claude Code / Codex subscription into a Feishu agent your whole team can @.

How to install Larkway

This entry records only its repository, not the path inside it, so there is no exact command to give. Open chuckwu0/larkway and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Larkway does

Turn your Claude Code / Codex subscription into a Feishu agent your whole team can @.

Alternatives in Development

Understand Anything — Graphs that teach graphs that impress 80.7k ★
Understand Anything — Claude Code skills that turn any codebase into an interactive knowledge graph you can explore, search, and ask 2.5k ★
Beautiful Feishu Whiteboard — 35 curated colour palette styles for building beautiful, editable Feishu / Lark (飞书) whiteboards 687 ★

README

Larkway

Turn your Claude Code or Codex subscription into a Feishu (Lark) agent your whole team can @-mention.

🌐 [larkway.dev](https://larkway.dev) · [中文版](README.zh.md)

You @ the bot in a Feishu thread. It runs on your machine — reading your real codebase, executing commands, opening MRs — and posts the result back. Connect a workspace that already works with Claude Code or Codex, or create a managed one. Larkway connects Feishu triggers to that local runtime.

**Current release: v0.3.73**

💬 Join the community (Feishu)

See Larkway in action and talk through how to set it up for your team — scan to join the **Larkway 交流群** (QR never expires):

Why

The engineers on your team know how a feature is actually implemented. Everyone else has to ask them. That friction compounds: a data analyst wondering whether a metric counts deleted records, a PM asking which API endpoint powers a chart, a QA engineer trying to reproduce a bug without context.

A generic AI assistant cannot answer these questions reliably — it has no access to your code. A bot that runs Claude Code or Codex against your real repository can.

Larkway makes that bot trivially installable and persistent: one `larkway start` on an always-on machine, one `@mention` in any Feishu thread, and your team has a self-serve answer to "how is this actually implemented."

How it works

Feishu thread
  │  @bot "how does the retention metric handle deleted users?"
  │
  ▼  WebSocket long-connection (outbound only, no public endpoint needed)
larkway bridge
  │  spawns subprocess  ──►  claude --resume  -p ""
  │                          (or: codex ...)
  │
  ▼  stream-json (NDJSON)
lark.card                ◄──  agent reads repo, runs bash, edits files,
  │  throttled PATCH           opens MR, starts dev preview
  │
  ▼