Skillv1.0.2

ClawScan security

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

Scanner verdict

ReviewMar 11, 2026, 12:40 PM

Verdict: Review
Confidence: medium
Model: gpt-5-mini
Summary: The skill's code and instructions generally match its stated purpose (scrape A-share Top10 and give buy/hold analysis), but there are omissions and small inconsistencies (missing dependency/install instructions, directory-name typos, and a requirement to verbatim-paste script output) that warrant caution before installing or running it.
Guidance: What to consider before installing/running: - Dependencies: The script uses Python, pandas, and Playwright (which may download Chromium). The skill has no install instructions; plan to install these in a controlled environment (virtualenv or container). - Run in isolation: Because Playwright will launch a browser and perform network requests, run the skill in a sandboxed environment (container or VM) to avoid unexpected side effects or large downloads. - Review and test: Read the script (already included) and run it manually first to confirm it only prints the expected table and messages. Verify no hidden network endpoints or unexpected stdout are produced in your environment. - Verbatim output requirement: SKILL.md demands pasting the script's entire raw terminal output into the agent response. This is fine now since the script prints only market data, but be cautious: if the script is modified later, this behavior could leak unexpected information. Prefer to inspect output locally before allowing the agent to reproduce it remotely. - Directory-name typo: SKILL.md references both 'achaser' and 'astockchaser' paths; actual file is scripts/list_top10.py at the repo root. Ensure you run the correct path to avoid confusion. - Trust & provenance: The source/homepage is unknown. If you will rely on this for decision-making, prefer code from a known author or fork and harden dependencies. If you need help creating a safe install/run plan (requirements.txt, virtualenv, or container), I can provide step-by-step instructions.

Review Dimensions

Purpose & Capability: noteName/description (A 股 Top10 + buy/hold analysis) aligns with the provided Python script which scrapes data from data.10jqka.com.cn and prints a Top10 table. The use of Playwright for scraping is reasonable for a dynamic site, but the skill does not declare that heavy dependency (playwright + browser binaries + pandas) in its metadata or install spec.
Instruction Scope: noteSKILL.md instructs the agent to run the included script and to paste the script's entire raw terminal output verbatim before analysis. The script itself only prints scraping progress and a DataFrame of market rows (no access to local files or env vars). Requiring verbatim output could unintentionally surface unexpected runtime data if the script is modified or the environment prints extra info; the instruction is strict but currently consistent with the script's visible behavior.
Install Mechanism: concernThere is no install spec even though the script requires third-party packages (playwright, pandas) and a Chromium browser. Playwright typically needs browser binaries to be installed/downloaded; absence of installation guidance is a practical omission and increases friction and risk (unexpected network downloads at runtime). This is disproportionate to an instruction-only metadata entry and should have explicit dependency and install steps.
Credentials: okThe skill requests no environment variables, no credentials, and does not access config paths. The network access in the script is limited to scraping a single public market data page (https://data.10jqka.com.cn/market/zdfph/). There are no obvious unrelated credentials being requested or accessed.
Persistence & Privilege: okFlags show always:false and no special privileges. The skill does not request persistent presence or modify other skills/configuration. Autonomous invocation is allowed by default but is not combined with other high-risk factors here.