{
  "id": "page:docs-adapters-reference-agents-qwen",
  "_kind": "Page",
  "_file": "wiki/docs/adapters/reference/agents/qwen.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/adapters/reference/agents/qwen.md",
    "sourceKind": "repo-docs",
    "title": "Qwen Code",
    "displayName": "Qwen Code",
    "slug": "docs/adapters/reference/agents/qwen",
    "articlePath": "wiki/docs/adapters/reference/agents/qwen.md",
    "article": "\n# Qwen Code\n\n[`@qwen-code/qwen-code`](https://github.com/QwenLM/qwen-code) is Alibaba's\ncoding CLI built on top of Gemini CLI, tuned for the Qwen3-Coder family of\nmodels. adapters drives it via the `qwen` binary.\n\n## Install\n\n```bash\nadapters install qwen\n```\n\nSupported on macOS, Linux and Windows.\n\n## Authenticate\n\nQwen Code speaks the OpenAI-compatible API. The quickest path is DashScope:\n\n```bash\nexport OPENAI_API_KEY=sk-...\nexport OPENAI_BASE_URL=https://dashscope-intl.aliyuncs.com/compatible-mode/v1\nexport OPENAI_MODEL=qwen3-coder-plus\n```\n\nAlternatively, run `qwen` interactively and use the built-in OAuth flow.\n\n## Example\n\n```ts\nimport { createClient } from '@a5c-ai/adapters';\n\nconst client = createClient();\nconst run = client.run({\n  agent: 'qwen',\n  model: 'qwen3-coder-plus',\n  prompt: 'Summarize the public API in src/index.ts',\n});\n\nfor await (const ev of run.events()) {\n  if (ev.type === 'text_delta') process.stdout.write(ev.delta);\n}\n```\n\n## Plugins\n\nPlugin support: **no**. Use MCP servers for extensibility.\n\n### MCP Servers\n```bash\nadapters mcp install qwen <mcp-server>\nadapters mcp list qwen\n```\n\nRegistry: https://modelcontextprotocol.io\n\n## Notes\n\n- Session files live under `~/.qwen/sessions` (JSONL).\n- MCP servers are configured under `mcpServers` in `~/.qwen/settings.json`.\n- Capabilities are set conservatively — thinking, JSON mode, and image input\n  default to `false` pending upstream confirmation.\n- Qwen Code 0.16.2 keeps the npm package unchanged. Upstream adds\n  `.qwen/QWEN.local.md` project-local context, the `memory-leak-debug` skill,\n  background-agent concurrency limits, Token Plan cache control, a three-tier\n  auto-compaction ladder, headless runaway-protection guardrails, startup\n  `--worktree` support, and SDK fixes for `canUseTool` timeouts and packaged\n  CLI chunks. The graph tracks these as catalog/runtime compatibility metadata;\n  adapters does not yet expose a typed worktree option for Qwen launch.\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-adapters-reference",
      "to": "page:docs-adapters-reference-agents-qwen",
      "kind": "contains_page"
    }
  ]
}