ClawHub

SwanLab Reader

v1.1.2

读取 SwanLab 实验数据。当用户发来 SwanLab run URL 时使用。

⭐ 0· 138·1 current·1 all-time
by@ckqiao

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for ckqiao/swanlab-reader.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "SwanLab Reader" (ckqiao/swanlab-reader) from ClawHub.
Skill page: https://clawhub.ai/ckqiao/swanlab-reader
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Required env vars: SWANLAB_KEY, SWANLAB_API_KEY
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install swanlab-reader

ClawHub CLI

Package manager switcher

npx clawhub@latest install swanlab-reader
Security Scan
VirusTotalVirusTotal
Pending
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
Name/description, declared dependencies (swanlab, numpy), and the code all target SwanLab experiment data. No unrelated services, binaries, or credentials are requested.
✓
Instruction Scope
SKILL.md commands map directly to the Python implementation. Instructions only read/write the skill's config key file (~/.config/swanlab/key) and call SwanLab endpoints or the swanlab library; they do not reference unrelated system files, credentials, or external endpoints.
✓
Install Mechanism
There is no installer that downloads arbitrary code; this is an instruction+source bundle that relies on the listed Python dependencies. The SKILL.md notes using 'uv run' to manage dependencies, which is proportionate to the task.
✓
Credentials
No required environment variables are enforced. The optional SWANLAB_KEY / SWANLAB_API_KEY and a local config file (~/.config/swanlab/key) are appropriate for an API client. No other secrets or unrelated env vars are requested.
✓
Persistence & Privilege
The skill is not marked always:true and does not modify other skills or global agent settings. It stores an API key in its own config directory (~/.config/swanlab), which is expected behavior for a CLI/API client.
Assessment
This skill appears to do what it says: query SwanLab runs and display metrics. Before installing, be aware it will: (1) contact https://swanlab.cn to fetch data; (2) optionally store any API key you set in plaintext at ~/.config/swanlab/key (or you can provide it via SWANLAB_KEY or SWANLAB_API_KEY environment variables); and (3) require the swanlab Python package (review that package if you want to validate its behavior). Do not provide unrelated credentials. If you later suspect misuse, revoke the SwanLab API key and remove the ~/.config/swanlab/key file.

Like a lobster shell, security has layers — review code before you run it.

Runtime requirements

Environment variables
SWANLAB_KEYrequired
SWANLAB_API_KEYrequired
latestvk97aq7r1rp1mvkemgg8xcfbrex83n2q6
138downloads
0stars
4versions
Updated 1mo ago
v1.1.2
MIT-0
@ckqiao
latest
Download

SwanLab Reader

安装

# 依赖：swanlab、numpy（uv run 自动处理）
uv run python skills/swanlab-reader/swanlab_reader.py set-key <your-key>

首次设置 API Key（只需一次）。key 保存在 ~/.config/swanlab/key，或通过环境变量 SWANLAB_KEY / SWANLAB_API_KEY 传入。

命令

runs — 项目概览

uv run python skills/swanlab-reader/swanlab_reader.py runs <username/project>

info — 指标 + 配置

uv run python skills/swanlab-reader/swanlab_reader.py info <url>

末尾输出可直接复制粘贴的 history 命令。

history — 历史序列

uv run python skills/swanlab-reader/swanlab_reader.py history <url> <key1> [key2] ...
uv run python skills/swanlab-reader/swanlab_reader.py history <url> <key> --plot   # 带 ASCII 图

plot — 单独画图

uv run python skills/swanlab-reader/swanlab_reader.py plot <url> <key>

compare — 多 run 对比

uv run python skills/swanlab-reader/swanlab_reader.py compare <url1> <url2> [url3 ...]

自动获取所有 run 的所有指标并列对比。

工作流

  1. runs <project> 看项目下有哪些 run
  2. info <url> 看指标列表 + 配置，复制末尾的 history 命令
  3. history <url> <keys> 查具体指标历史
  4. compare <url1> <url2> ... 多 run 并排对比

Comments

Loading comments...