← All skills
clawevolve-bench
90 / 100 · Low risk
Fine to try. Some of what it can do is expected: a devops skill usually runs commands, installs tools and uses cloud credentials.
By inclusionAI/Avernet (GitHub) · devops · 602 stars · scanned at commit f978637
以 Python workflow 方式完整运行一次 ClawWeb Bench，不依赖 ClawMind workflow；可由 ClawEvolve Step handler 或 Optimize 调用。
Works with any agent that supports SKILL.md skills.
What this skill can do
Runs commands on your computer
The author does not mention this. Expected for a devops skill.
Found in 17 places.
Show 14 more
- apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/scripts/clawbench-workflow.py, line 371
command_runner: Callable[..., subprocess.CompletedProcess[str]] = subprocess.run, - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/scripts/clawbench-workflow.py, line 763
except subprocess.TimeoutExpired as exc: - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 80
completed = workflow_module.subprocess.CompletedProcess( - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 210
runner = mock.Mock(return_value=workflow_module.subprocess.CompletedProcess( - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 224
runner = mock.Mock(return_value=workflow_module.subprocess.CompletedProcess( - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 247
return workflow_module.subprocess.CompletedProcess(argv, 0, "- report-agent\n", "") - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 252
return workflow_module.subprocess.CompletedProcess(argv, 0, json.dumps(payload), "") - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 296
return workflow_module.subprocess.CompletedProcess(argv, 0, "", "") - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 298
return workflow_module.subprocess.CompletedProcess(argv, 0, json.dumps(payload), "") - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 330
return workflow_module.subprocess.CompletedProcess(argv, 0, "- report-agent\n", "") - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 331
return workflow_module.subprocess.CompletedProcess( - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 357
return workflow_module.subprocess.CompletedProcess(argv, 0, "", "") - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 359
return workflow_module.subprocess.CompletedProcess(argv, 0, "created", "") - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 360
return workflow_module.subprocess.CompletedProcess(
Connects to the internet
The author does not mention this. Expected for a devops skill.
Found in 8 places.
Show 5 more
- apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 138
with mock.patch.object(workflow_module.urllib.request, "urlopen", return_value=response) as urlopen: - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 142
payload = json.loads(urlopen.call_args.args[0].data.decode("utf-8")) - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 153
with mock.patch.object(workflow_module.urllib.request, "urlopen", side_effect=error): - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 166
with mock.patch.object(workflow_module.urllib.request, "urlopen", return_value=response) as urlopen: - apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench/tests/test_clawbench_workflow.py, line 171
request = urlopen.call_args.args[0]
Install
Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.
if [ -e ~/.agents/skills/'clawevolve-bench' ] || [ -L ~/.agents/skills/'clawevolve-bench' ]; then echo 'A skill folder named clawevolve-bench already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/inclusionAI/Avernet' "$d" && git -C "$d" checkout --quiet 'f97863707dfe9c4d140b8b39cfc0564fa168627d' -- 'apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench' && mkdir -p ~/.agents/skills && cp -R "$d"/'apps/evolverun/clawweb-skills/clawevolve-skills/clawevolve-bench' ~/.agents/skills/'clawevolve-bench' && rm -rf "$d"; fi
Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'clawevolve-bench' ~/.claude/skills/'clawevolve-bench' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.
Badge for your README
The badge for inclusionAI/Avernet shows the score of its lowest-scoring skill.
[](https://zalavijaysinh.github.io/skilltrust/repos/inclusionAI/Avernet/)
Score history
| Date | Version | Score |
|---|
| 2026-11-12 | f978637 | 90 |
| 2026-11-08 | 58db87b | 90 |
| 2026-11-06 | 65e270d | 90 |