How do I install the Qwen Harness Tools skill?

Run npx skillmds@latest add ariffazil/qwen-harness-tools in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Qwen Harness Tools safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: makes network calls, reads secrets. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Qwen Harness Tools?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Qwen Harness Tools free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Qwen Harness Tools?

ariffazil (@ariffazil) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Qwen Harness Tools by ariffazil — Agent Skill

Name: Qwen Harness Tools
Rating: 4.2 (5 reviews)
Author: ariffazil

Qwen Harness Tools

Documents Qwen Token Plan Harness tools (web search, code interpreter, web scraping, reverse image search, text-to-image search) that are built into supported models. Activates when the user asks for real-time web info, code execution in a sandbox, or image search while using a Token Plan model.

ariffazil Updated Sep 10, 2026

File contents

Qwen Token Plan Harness tools are built into supported models. The model invokes them automatically based on the question — no extra configuration, billed per successful invocation from Token Plan Credits.

User request: $ARGUMENTS

Applicable models (Personal + Team Edition)

Model	Harness tools
`qwen3.8-max`	web search, code interpreter, web scraping, reverse image search, text-to-image search
`qwen3.7-max`	web search, code interpreter, web scraping
`qwen3.7-plus`	web search, code interpreter, web scraping, reverse image search, text-to-image search

Note: Harness tools apply to Token Plan, not to Coding Plan.

Tool reference

Tool	What it does	When the model uses it
Web search	Retrieves information from the internet and generates answers based on search results.	Questions about current events, recent releases, news, prices, etc.
Code interpreter	Writes and runs Python code in a sandbox for math / data analysis.	Numerical computation, data transformation, plotting, statistical analysis.
Web scraping	Accesses a specified URL and extracts its content.	User gives a URL and asks for info from that page.
Reverse image search	Searches for visually similar images given an input image.	"Find where this image came from", "find similar products".
Text-to-image search	Searches for relevant images given a text description.	"Find me an image of X", "what does X look like".

Steps

Verify the active model is in the supported list above. If not, switch via the tool's model picker:
- Claude Code: /model qwen3.7-max
- OpenCode: /models then select
- Qwen Code: /model then select
- OpenClaw: update ~/.openclaw/openclaw.json agents.defaults.model.primary
Just ask. Harness tools are automatic — the model decides when to invoke based on the question. Example: Use the qwen3.7-max built-in web search to find the latest on <topic>.
For URL scraping specifically, include the URL in the prompt so the model routes to the web scraping tool (not web search):

Use the web scraping tool on https://example.com/page and summarize.
For code interpreter, ask computational questions explicitly:

Use code interpreter to compute the standard deviation of <list>.
Cost control: Harness invocations are billed per successful call. Avoid invoking on trivial questions where the model could answer from context alone.

Notes

Token Plan only: these tools do NOT apply to Coding Plan, pay-as-you-go API keys (sk- prefix), or FLAME RM0-TOOLS-FREELOOP chain.
Vision vs Harness: separate capability. Vision (image input) is auto-enabled for qwen3.8-max, qwen3.7-plus, qwen3.6-plus, qwen3.5-plus, kimi-k2.5. Harness is a tool, not vision.
Free alternative: for RM0 web search outside Token Plan, use the firecrawl-web-search skill instead.
OpenCode / OpenClaw config: see token-plan-bailian-config skill for the modalities / input fields that enable vision.

---name: qwen-harness-toolsdescription: Documents Qwen Token Plan Harness tools (web search, code interpreter, web scraping, reverse image search, text-to-image search) that are built into supported models. Activates when the user asks for real-time web info, code execution in a sandbox, or image search while using a Token Plan model.---Qwen Token Plan **Harness tools** are built into supported models. The model invokes them automatically based on the question — no extra configuration, billed per successful invocation from Token Plan Credits.User request: $ARGUMENTS## Applicable models (Personal + Team Edition)| Model | Harness tools ||---|---|| `qwen3.8-max` | web search, code interpreter, web scraping, reverse image search, text-to-image search || `qwen3.7-max` | web search, code interpreter, web scraping || `qwen3.7-plus` | web search, code interpreter, web scraping, reverse image search, text-to-image search |**Note:** Harness tools apply to **Token Plan**, not to **Coding Plan**.## Tool reference| Tool | What it does | When the model uses it ||---|---|---|| **Web search** | Retrieves information from the internet and generates answers based on search results. | Questions about current events, recent releases, news, prices, etc. || **Code interpreter** | Writes and runs Python code in a sandbox for math / data analysis. | Numerical computation, data transformation, plotting, statistical analysis. || **Web scraping** | Accesses a specified URL and extracts its content. | User gives a URL and asks for info from that page. || **Reverse image search** | Searches for visually similar images given an input image. | "Find where this image came from", "find similar products". || **Text-to-image search** | Searches for relevant images given a text description. | "Find me an image of X", "what does X look like". |## Steps1. **Verify the active model is in the supported list above.** If not, switch via the tool's model picker: - Claude Code: `/model qwen3.7-max` - OpenCode: `/models` then select - Qwen Code: `/model` then select - OpenClaw: update `~/.openclaw/openclaw.json` `agents.defaults.model.primary`2. **Just ask.** Harness tools are automatic — the model decides when to invoke based on the question. Example: `Use the qwen3.7-max built-in web search to find the latest on <topic>.`3. **For URL scraping specifically**, include the URL in the prompt so the model routes to the web scraping tool (not web search): > `Use the web scraping tool on https://example.com/page and summarize.`4. **For code interpreter**, ask computational questions explicitly: > `Use code interpreter to compute the standard deviation of <list>.`5. **Cost control**: Harness invocations are billed per successful call. Avoid invoking on trivial questions where the model could answer from context alone.## Notes- **Token Plan only**: these tools do NOT apply to Coding Plan, pay-as-you-go API keys (`sk-` prefix), or FLAME `RM0-TOOLS-FREELOOP` chain.- **Vision vs Harness**: separate capability. Vision (image input) is auto-enabled for `qwen3.8-max`, `qwen3.7-plus`, `qwen3.6-plus`, `qwen3.5-plus`, `kimi-k2.5`. Harness is a tool, not vision.- **Free alternative**: for RM0 web search outside Token Plan, use the `firecrawl-web-search` skill instead.- **OpenCode / OpenClaw config**: see `token-plan-bailian-config` skill for the modalities / input fields that enable vision.

Frequently asked questions