Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
aliyun avatar

Alibabacloud Agentbay Aio Skills

  • 112 installs
  • 208 repo stars
  • Updated August 4, 2026
  • aliyun/alibabacloud-aiops-skills

AgentBay AIO is a Claude Code skill that executes user code in a secure remote Alibaba Cloud sandbox via the AgentBay SDK.

About

AgentBay AIO is a skill that runs user code in a secure remote Alibaba Cloud sandbox via the AgentBay SDK instead of on the local machine. A developer uses it whenever they ask the agent to run, execute, plot, or evaluate code in Python, JavaScript, R, or Java. It handles sandbox creation, execution, and result parsing through a single entry script and validates downloaded files like chart images.

  • Executes Python, JavaScript, R, and Java in a remote sandbox, never locally
  • Handles chart plotting with automatic CJK font configuration for matplotlib
  • 60-second execution timeout with structured JSON output mode

Alibabacloud Agentbay Aio Skills by the numbers

  • 112 all-time installs (skills.sh)
  • Ranked #548 of 1,039 Cloud & Infrastructure skills by installs in the Skillselion catalog
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
At a glance

alibabacloud-agentbay-aio-skills capabilities & compatibility

Requires an AgentBay API key from the AgentBay Console; sandbox usage billed by Alibaba Cloud.

Capabilities
data analysis
Use cases
data analysis
Platforms
macOS · Linux · Windows
Pricing
Bring your own API key
From the docs

What alibabacloud-agentbay-aio-skills says it does

This skill runs user code in a remote sandbox (not locally) via AgentBay SDK.
SKILL.md
Supported languages: `python`, `javascript`, `r`, `java` (case-insensitive). Timeout: 60 seconds max (`--timeout-s`).
SKILL.md
npx skills add https://github.com/aliyun/alibabacloud-aiops-skills --skill alibabacloud-agentbay-aio-skills

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs112
repo stars208
Last updatedAugust 4, 2026
Repositoryaliyun/alibabacloud-aiops-skills

What it does

Run or evaluate user code (Python, JS, R, Java) in a secure remote Alibaba Cloud AgentBay sandbox instead of locally.

Who is it for?

Running or evaluating untrusted or ad-hoc code and plotting charts without executing anything on the local machine.

Skip if: Tasks the agent should run directly in the local terminal; the skill forbids local execution as a fallback.

When should I use this skill?

The user asks to run, execute, or evaluate code, plot with Python, or run data analysis.

What you get

User code runs in an isolated cloud sandbox with parsed results and validated file downloads.

By the numbers

  • 4 supported languages (python/javascript/r/java)
  • 60-second max timeout

Files

SKILL.mdMarkdownGitHub ↗

AgentBay AIO Skill

How It Works

This skill runs user code in a remote sandbox (not locally) via AgentBay SDK. The entry script scripts/run_code.py handles sandbox creation, code execution, and result parsing automatically.

# Basic execution
python scripts/run_code.py --code "<user_code>" --language python

# Execute from file
python scripts/run_code.py --code-file /path/to/file.py --language python

# Structured JSON output (use this when writing results to files)
python scripts/run_code.py --code "<user_code>" --language python --json

Supported languages: python, javascript, r, java (case-insensitive). Timeout: 60 seconds max (--timeout-s).

Prerequisites

DependencyVersionPurpose
Python>= 3.8Runtime
wuying-agentbay-sdk>= 1.0.0AgentBay SDK for sandbox code execution

Setup

The script requires wuying-agentbay-sdk installed locally. Install it before first use:

pip install wuying-agentbay-sdk

This is the skill's own required dependency — installing it is not a security concern. If run_code.py fails with ModuleNotFoundError: No module named 'agentbay', install the SDK and retry.

Network domains used: agentbay.aliyuncs.com (API endpoint), agentbay.console.aliyun.com (console), mirrors.aliyun.com (PyPI mirror).

API Key Configuration

Run scripts directly without prompting for API Key configuration. Only guide users when the script explicitly reports "Missing API key":

1. Apply at AgentBay Console 2. Save to config file: ~/.config/agentbay/api_key (macOS/Linux) or %USERPROFILE%\.config\agentbay\api_key (Windows)

Execution Rules

CRITICAL: All user code MUST run through scripts/run_code.py — NEVER execute code directly in the local terminal (e.g., python -c, node -e, timeout ... python). This applies even if run_code.py fails on the first attempt.

  • If run_code.py fails with a transient error, retry once before reporting failure.
  • If run_code.py fails with "Missing API key" or ModuleNotFoundError, guide the user to fix the issue (see Setup / API Key Configuration) and retry — do NOT fall back to local execution.
  • NEVER run user code locally as a fallback. Report the error instead.

Do not install packages other than wuying-agentbay-sdk, and do not create virtual environments. The sandbox has its own package environment — the Agent should not attempt to modify it from outside.

Output Handling

Standard output: Exit code 0 = success, results in stdout. Non-zero = failure, error in stderr.

Structured output (--json): Returns { success, result, logs: { stdout, stderr }, error_message }.

Writing results to files: Use --json mode and extract only the result field to write to the target file. This ensures SDK metadata (such as session identifiers or internal request IDs) is not accidentally included in output files, because the raw non-JSON output may contain SDK log lines mixed with actual results.

Reporting results: Quote the script's original output directly. Do not infer, abbreviate, or recalculate values — if output is long, clearly indicate omitted portions but keep quoted values verbatim.

File Download Validation

When saving files generated in the sandbox (e.g., chart images) to the local environment:

1. Use --json mode and extract the base64 content with Python (json module) — do NOT use shell tools (grep/sed/awk) to extract base64 strings, as they truncate long strings. 2. Decode with: python -c "import base64,json,sys; d=json.load(sys.stdin); open('out.png','wb').write(base64.b64decode(d['result']))" < output.json 3. Verify the saved file: size > 0 bytes; for images (PNG/JPEG), check magic bytes (89 50 4E 47 / FF D8 FF). 4. For chart images, file size should be > 5 KB. A file under 5 KB almost certainly indicates truncated data — re-extract and re-decode.

If verification fails, retry the download. Do not report success with a corrupted file.

Chinese/CJK Character Rendering

The script (run_code.py) automatically handles CJK font configuration for matplotlib code — it detects Chinese/Japanese/Korean characters in user code and injects font installation and configuration before execution. No manual font setup is needed.

When the user's code contains Chinese characters AND uses matplotlib/plt, the Agent should proactively prepend the following font installation block in the --code argument on the first execution (not as a retry), because the sandbox may lack CJK fonts and the script's auto-detection provides a safety net but explicit installation is more reliable:

import subprocess
subprocess.run(['apt-get', 'update', '-qq'], capture_output=True)
subprocess.run(['apt-get', 'install', '-y', '-qq', 'fonts-wqy-microhei'], capture_output=True)
import matplotlib
import matplotlib.font_manager as fm
fm.fontManager.addfont('/usr/share/fonts/truetype/wqy/wqy-microhei.ttc')
fm.fontManager = fm.FontManager()
matplotlib.rcParams['font.family'] = 'WenQuanYi Micro Hei'
matplotlib.rcParams['axes.unicode_minus'] = False

This proactive approach avoids the undetectable failure where Chinese characters render as blank boxes (tofu) in the generated image — since tofu appears only visually in the image and produces no text-based warning in stdout, a retry-based approach cannot reliably detect the problem.

Information Security

Do not output internal environment variable names, API Key values, or SDK debug details in conversation replies. When executing scripts, use --json mode and display only the result and error_message fields to users. The raw SDK output may contain internal fields (session identifiers, request IDs, access credentials) that should not be exposed to users or written to output files.

Related skills

FAQ

Which languages does AgentBay AIO support?

Python, JavaScript, R, and Java, case-insensitive, with a 60-second maximum timeout.

Does it ever run code locally?

No. It always runs code through the remote sandbox and never falls back to local execution.

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.