Skillv1.0.0

ClawScan security

Deepseek Extract · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignApr 21, 2026, 12:16 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's code, instructions, and requirements are consistent with its stated purpose (extracting DeepSeek shared chat pages); it does not request unrelated credentials or perform obvious exfiltration.
Guidance: This skill appears to do what it says: locally render a DeepSeek share URL with Playwright and extract messages. Before installing or running it, note: (1) you will likely need to run pip install playwright and playwright install chromium, which downloads browser binaries into your environment; (2) the script opens and executes JavaScript on the target page (the provided DeepSeek URL) to render content — only provide share URLs you trust or are authorized to extract; (3) the skill does not request credentials or exfiltrate data to third parties, but installing Python packages always carries the usual supply-chain risk, so install packages from trusted sources and inspect code if you have concerns; (4) the URL validation is strict and may reject some valid share URLs with extra query strings or atypical formats. If you want greater assurance, review the full script contents locally before running and run Playwright installs in an isolated environment (virtualenv or container).

Review Dimensions

Purpose & Capability: okName/description match the included script and SKILL.md: the skill uses Playwright + headless Chromium to render DeepSeek share pages and extract messages. Required binary is just python3, which is reasonable for a Python extraction script. Nothing requested (no credentials, no unrelated binaries) appears out of scope for the stated task.
Instruction Scope: okSKILL.md restricts usage to chat.deepseek.com/share/... URLs and instructs running the included script to produce Markdown/JSON output. The runtime instructions ask only to run the local script, check for Playwright, read the output file, and optionally retry with headed mode; they do not direct the agent to read unrelated local files, other credentials, or post data to third-party endpoints. The extraction code navigates only to the provided URL (after validating it) and parses page content.
Install Mechanism: noteThis is an instruction-only skill with a bundled Python script (no automated install spec). It requires the user to install Playwright and its Chromium browsers (pip install playwright; playwright install chromium). Installing Playwright will download browser binaries from the internet; this is expected for a browser-based scraper but is a worthwhile operational note for users (it modifies local environment and adds large browser binaries).
Credentials: okThe skill requires no environment variables, no credentials, and no config paths. The code does not attempt to access system secrets or other services beyond loading the provided DeepSeek URL.
Persistence & Privilege: okThe skill is not always-on and does not request persistent privileges. It does not modify other skills' configs or system-wide agent settings. The agent may run it autonomously (default), which is normal and not excessive here.