Qwen Agent

Offloads mechanical coding work to a cheaper Qwen model so you don't burn Claude tokens on grunt work. You point it at bulk renames, formatting passes, boilerplate generation, or test scaffolding with a self-contained prompt and absolute paths, and it runs headless through the claude-9arm alias. The big gotcha is Qwen's 128k context window, so you need to chunk large jobs into per-file or per-directory slices. Works well when the task is boring and low-risk. Don't use it for anything that needs design judgment, debugging reasoning, or context from your conversation. Verify the output yourself since cheaper models make cheaper mistakes.

skills.sh installs: 2.5K
GitHub stars: 3.3K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add thananon/9arm-skills --skill qwen-agent --agent claude-code

Installs into .claude/skills of the current project.

Files

qwen-agent

Offload menial, self-contained tasks to a Qwen model running inside a headless Claude Code instance (claude-9arm). Keeps expensive Claude reasoning for work that needs it.

The command

claude-9arm is a shell alias → claude --model qwen3.6-35b-a3b routed through the 9arm gateway. Run it headless with -p:

claude-9arm -p "<self-contained task prompt>" --allowedTools Bash Read Edit Write Glob Grep

This is the default invocation. The flag list scopes which tools the subagent may use without a prompt, so it can finish a menial job unattended. Without it the subagent stalls waiting for approval on the first edit or command.
The alias bakes in --allowedTools '*', which Claude Code silently ignores with a warning (Wildcard tool name "*" is not supported). That warning is expected and harmless — the --allowedTools you append is what takes effect.
For edit-only, lower-risk tasks you may instead use --permission-mode acceptEdits (auto-accepts file edits, but Bash still prompts — don't use it for verification/build/test runs).