← Back to Skills Marketplace
241
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install media-cluster
Description
Automatically crawls Chinese social media by keyword, summarizes content, generates a markdown report, and produces a short voice summary using TTS.
Usage Guidance
This skill mostly does what it says (clones a MediaCrawler repo, runs crawls, summarizes, and optionally calls SenseAudio for TTS), but you should be careful before installing/running it: 1) Metadata is incomplete — the skill requires Conda, Node.js for some platforms, and an API key SENSEAUDIO_API_KEY for TTS, but those are not declared in the registry entry; ask the author to correct the metadata. 2) The skill clones and executes a third‑party GitHub repository at runtime — review that repository's code yourself (or run inside an isolated VM/container) before allowing it to run on any machine with sensitive data. 3) Running crawlers may require logging into target apps (QR scan) and may store login state; consider privacy and terms‑of‑service/legal issues for scraping. 4) If you only need summarization/TTS of user-supplied content, consider using an approach that does not run a remote crawler. 5) If you proceed, run the skill in an isolated environment (container/VM) and provide a SenseAudio API key only if you trust the TTS provider; revoke keys if you suspect misuse.
Capability Analysis
Type: OpenClaw Skill
Name: media-cluster
Version: 1.0.1
The skill bundle automates social media crawling and summary generation by cloning the external 'MediaCrawler' repository and executing its code via shell scripts (ensure_mediacrawler.sh, setup_env.sh), which introduces a significant supply chain and RCE risk. The SKILL.md file contains instructions for the agent to automatically execute a multi-step workflow—including environment setup, third-party code execution, and API interaction—upon a single user prompt, which increases the potential impact of prompt injection. While these capabilities (shell access, network calls to https://api.senseaudio.cn, and file system operations) are aligned with the stated purpose, the reliance on unpinned external artifacts and automated high-privilege execution steps warrants a suspicious classification.
Capability Assessment
Purpose & Capability
The skill's name/description (crawl Chinese social media, summarize, TTS) align with the included scripts: it clones a MediaCrawler repository, runs the crawler, summarises output, and calls SenseAudio for TTS. However the registry metadata declares no required env vars or binaries while the SKILL.md and scripts require Conda, Node.js (for some platforms), and a SENSEAUDIO_API_KEY for TTS — an inconsistency between description/behavior and declared requirements.
Instruction Scope
SKILL.md gives explicit commands: clone GitHub repo, create/activate conda env, pip install requirements, run the crawler, and run summarize_and_voice.py which posts to SenseAudio. All instructions stay within the stated purpose. Important operational behaviors: the agent will (if run) download a third‑party project and execute its code, open browsers for QR login, and may prompt for or persist login state. The instructions do not ask the agent to read unrelated system secrets, but they do require network access and executing external code.
Install Mechanism
There is no install spec in the registry; at runtime the skill clones https://github.com/NanmiCoder/MediaCrawler.git and installs packages via pip/playwright and creates a conda env. Downloading and executing code from an external GitHub repo is expected for this functionality but increases risk because arbitrary third‑party code will run on the host. The download URL is a normal GitHub repo (not a shortened/personal server), which is better than an untrusted host, but still introduces execution-of-remote-code risk.
Credentials
SKILL.md and scripts require Conda, Node.js (>=16 for some crawls), and an API key SENSEAUDIO_API_KEY (and optional SENSEAUDIO_VOICE_ID). The registry metadata lists no required env vars or binaries — this mismatch is important: the skill will fail or behave differently without those, and the missing declaration means the platform/user consent step may not surface these requirements. No other unrelated credentials are requested.
Persistence & Privilege
The skill does not request 'always: true' and does not modify other skills' configs. It runs when invoked and may create a conda environment and cache login state for crawled platforms — normal for this function. Autonomous invocation is allowed (platform default) but is not combined with other high privileges here.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install media-cluster - After installation, invoke the skill by name or use
/media-cluster - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
**Improved cross-platform compatibility and simplified environment variable handling.**
- Updated documentation to provide separate setup instructions for Windows and macOS/Linux, improving clarity for all users.
- Voice API key and options are now read from environment variables instead of requiring a `.env` file.
- Removed the unused `.env` template file.
- Clarified that shell scripts are recommended for macOS/Linux and not for Windows; Windows users should follow manual steps.
- General documentation cleanup for conciseness and easier onboarding.
v1.0.0
- Initial release of the media-cluster skill.
- Enables one-sentence social media crawling and summarization for platforms like 小红书, 抖音, 微博, and B站 by keyword.
- Fully automated workflow: environment setup, crawling, report generation, and short voice summary.
- Uses MediaCrawler for data collection and SenseAudio for TTS voice summary.
- Users only need to input the query; the agent handles the rest automatically.
Metadata
Frequently Asked Questions
What is media-cluster?
Automatically crawls Chinese social media by keyword, summarizes content, generates a markdown report, and produces a short voice summary using TTS. It is an AI Agent Skill for Claude Code / OpenClaw, with 241 downloads so far.
How do I install media-cluster?
Run "/install media-cluster" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is media-cluster free?
Yes, media-cluster is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does media-cluster support?
media-cluster is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created media-cluster?
It is built and maintained by q1lin570 (@q1lin570); the current version is v1.0.1.
More Skills