Use MCP Servers to 10x Productivity (Simple, Practical)

MCP servers give Qwen superpowers: browse the web, run Playwright, talk to GitHub, recall memory, and search fast. Add a few, and you’ll plan, code, test, and research without leaving Qwen.

Setup (once)

Work inside 01_hello_qwen, then run qwen.
Keep secrets in your shell (export env vars), not in files.

Configuration basics (settings.json)

qwen mcp list → confirm servers
/mcp → see tools and auth status
Before configuring a new server: run qwen mcp list and /mcp to see baseline
File: .qwen/settings.json (project) or ~/.qwen/settings.json (global)

Recommended starters (what + how to add)

Playwright MCP (headless browser, capture, test flows)

Value: script user journeys; verify pages after changes.
Add .qwen/settings.json:

{
  "mcpServers": {
    "playwright": {
      "command": "npx",
      "args": [
        "@playwright/mcp@latest"
      ]
    }
  }
}

Configure Playwright as above, then run qwen mcp list and /mcp again to confirm
Ask (Playwright workflow prompt):
- "now setup a simple page that shows galaxy. Take the screenshot to ensure it's all perfect and review it. Continue iterating after reviewing to get perfect galaxy view"

Tavily Browser Search (web search + fetch)

Value: quick research, compare docs, gather refs.

{
  "mcpServers": {
    "tavily-remote": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.tavily.com/mcp/?tavilyApiKey=<your-api-key>"
      ]
    }
  }
}

Ask (search prompt): Check and research if gemini 3.0 and sonnet 4.5 are released.

Context7MCP - Up-to-date Code Docs For Any Prompt

Value: focused answers + sources for decisions.

{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp", "--api-key", "YOUR_API_KEY"]
    }
  }
}

OpenMemory (save facts across sessions)

    "openmemory": {
      "command": "npx",
      "args": [
        "-y",
        "openmemory"
      ],
      "env": {
        "OPENMEMORY_API_KEY": "YOUR_API_KEY"
      }
    }

Ask (mem prompt): Use add-memory tool and save that we prefer to use OpenAI Agents SDK, MCP, A2A, Kubernetes for Agentic AI.

Daily flow (2 minutes)

qwen mcp list → confirm servers
/mcp → see tools and auth status
Before configuring a new server: run qwen mcp list and /mcp to see baseline
Configure Playwright as above, then run qwen mcp list and /mcp again to confirm
Ask (Playwright workflow prompt):
- "now setup a simple page that shows galaxy. Take the screenshot to ensure it's all perfect and review it. Continue iterating after reviewing to get perfect galaxy view"
Save a team fact: SaveMemory("We release on Fridays 3pm UTC.")

Trust, safety, and approvals

Start with defaults (confirmation on tools). Toggle with /approval-mode when confident.
In .qwen/settings.json, you can whitelist/blacklist tools (e.g., allow Shell(git status), block rm).

How to choose servers (think like a technical business manager)

Outcome first: What decision or workflow does this unblock (e.g., “verify login flow daily”, “gather competitive notes in 5 minutes”)?

ROI quick check: Setup < 15 min, saves > 30 min/week, reduces context switching.
Security posture: Auth via env vars; no broad shell; whitelist tools; avoid storing tokens in repo.
Maintenance: Active repo, clear docs, simple add/remove commands, minimal deps.
Integration fit: Complements your stack (browser, search, memory); avoids overlap.
Proof-of-value: 1‑day pilot with a concrete metric (e.g., flaky login caught, 3 curated sources with quotes, meeting brief auto‑generated).

CLI Commands Subagents