Back to skill
Skillv1.0.3
ClawScan security
xtdata · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 23, 2026, 10:06 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's code and runtime instructions align with its stated purpose (providing XtQuant/miniQMT market-data access); it requests only python3 and no credentials, but it requires installing the xtquant package and connecting to a local miniQMT instance which can access local data files.
- Guidance
- This skill appears to do what it says, but review these points before installing: 1) pip install xtquant will fetch and execute third-party package code — verify the xtquant package source/version (PyPI/project homepage) if you need supply-chain assurance. 2) The library connects to a local miniQMT TCP service and may read the local userdata_mini path — only use it if you trust the local MiniQMT binary and the data files it will access. 3) No cloud credentials are requested by the skill itself, but network activity (PyPI install and local TCP) will occur. 4) Minor metadata/version mismatches exist between files (SKILL.md vs metadata), which looks like bookkeeping rather than malicious behavior. If you require higher assurance, inspect the xtquant package source and run installs in an isolated environment (virtualenv/container) before trusting it with production data.
Review Dimensions
- Purpose & Capability
- okName/description, documentation files, and the demo all describe a market-data module that communicates with a local miniQMT client. The declared requirement (python3) and the included docs/README/request to pip install xtquant are consistent with that purpose.
- Instruction Scope
- noteSKILL.md instructs the agent to pip install xtquant and to call xtdata.connect(), subscribe, download, and read local cache (get_local_data). This is within the module's purpose, but it explicitly requires connecting to a local TCP service (miniQMT) and reading a user data directory (userdata_mini), which means the skill will cause local network connections and read local data files when used.
- Install Mechanism
- noteThe skill is instruction-only (no install spec). SKILL.md and requirements.txt direct users to pip install xtquant and pandas — a normal approach but one that installs third-party packages from PyPI. Installing xtquant will run arbitrary code from that package, so verification of the upstream package (source, integrity) is advisable.
- Credentials
- okNo environment variables, credentials, or config paths are requested by the skill bundle. The only requested access is to the local miniQMT service and optionally a local data directory (data_dir), which is proportionate to a market-data library.
- Persistence & Privilege
- okalways is false and the skill does not request elevated platform privileges or persistent forcing into every agent run. Autonomous invocation is allowed (platform default) but not combined with broad credential requests.