Skillv2.0.0

ClawScan security

A股十书全景分析法 · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignApr 21, 2026, 1:51 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill is an instruction-only stock-analysis workflow that uses public finance webpages and internal reference files; its requirements and instructions are consistent with its stated purpose and it does not request credentials or install code.
Guidance: This skill appears internally consistent: it uses only public finance APIs and the included reference docs to produce analyses and backtests, and it does not ask for credentials or install code. Things to consider before enabling: (1) outputs are analyses not guaranteed advice — verify data and model conclusions before acting on them; (2) the skill fetches live data from third-party public endpoints (qt.gtimg.cn, sina finance, etc.); confirm those sources meet your accuracy/privacy requirements and your environment allows outbound HTTP to them; (3) backtest rules specify using unadjusted prices and only pre-cutoff financial reports — confirm that behavior matches your expectations for historical testing; (4) because the skill can be invoked by triggers and can call web_search, be mindful of what user-provided strings get sent to search services (sensitive/company-internal tickers or private notes may be exposed in queries). If you need stricter constraints (no web_search usage, whitelisted data sources, or audit logging), request those platform limits before installing.

Review Dimensions

Purpose & Capability: okName/description (A股十书全景分析法) align with what the skill does: it defines analysis paths, data sources, templates and backtest rules. It does not request credentials or unrelated system access that would be inconsistent with a stock-analysis skill.
Instruction Scope: okRuntime instructions are limited to: classify company type, fetch market/financial data from public finance endpoints, apply frameworks from bundled reference files, produce vote tables/ratings, and perform backtests using only pre-cutoff data. The instructions do not ask the agent to read local files, environment secrets, or system-level configuration outside the included reference docs. They do instruct web_search and HTTP fetches to public endpoints (expected for this purpose).
Install Mechanism: okNo install spec and no code files beyond instruction/reference markdowns. Because it's instruction-only, nothing is written to disk or downloaded at install time — lowest-risk install model.
Credentials: okThe skill requires no environment variables, credentials, or config paths. All external data access is to public finance endpoints and web search; requested accesses are proportional to a stock analysis tool and no unexplained secrets are needed.
Persistence & Privilege: okalways is false (not force-installed) and the skill does not request elevated or persistent system privileges or modify other skills. Autonomous invocation is allowed by platform default but not combined with other high-risk factors here.