Skillv1.0.3

ClawScan security

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

Scanner verdict

BenignMar 23, 2026, 10:05 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's files, instructions, and requirements are coherent with its stated purpose (a Python client that talks to a locally running miniQMT Windows service via the xtquant SDK); no obvious mismatches or hidden exfiltration were found, only a few small operational notes to consider.
Guidance: This skill appears to do what it says: it is a wrapper/docs for the xtquant SDK to connect to a locally running miniQMT Windows client for market data and trading. Before installing or running it: (1) confirm you actually have a miniQMT client and broker permission (the instructions expect Windows and a userdata_mini folder); (2) verify the xtquant package source (pip install xtquant) — prefer the official PyPI package or vendor-provided binaries and review its code if you have doubts; (3) be cautious when pointing the skill/agent to your userdata_mini path because it may contain sensitive broker configuration or credentials and allowing write/read access enables programmatic trading; (4) run first in an isolated/test environment (use a read-only copy of userdata or a test account) to confirm behavior; (5) if you need a stricter policy, restrict the agent so it cannot autonomously run trades or access your real userdata directory until manually authorized. If you can provide the actual xtquant package origin or a signed binary, I could raise confidence further; conversely, if xtquant were pulled from an untrusted URL rather than PyPI, that would increase risk.

Review Dimensions

Purpose & Capability: noteThe name/description, SKILL.md, docs, and demo all consistently describe using the xtquant Python SDK to connect to a locally running miniQMT (Windows) client for market data and trading. The declared requirement (python3) is appropriate. Minor mismatch: the documentation and examples assume a Windows miniQMT client and a Windows userdata_mini path, but the skill metadata does not restrict OS; this is an operational note rather than a security mismatch.
Instruction Scope: noteRuntime instructions tell users/agents to pip install xtquant and to connect to a local miniQMT service and to provide a path to the userdata_mini folder. Those steps are necessary for programmatic trading and data access. Important: the instructions require access to a local userdata directory (path) which may contain broker-related data/config; that access is expected for trading functionality but is sensitive and should be granted deliberately. No instructions attempt to call remote endpoints other than the documented xtquant APIs.
Install Mechanism: okThis is an instruction-only skill with no install spec; it suggests 'pip install xtquant' at runtime. No downloads from arbitrary URLs or archive extraction are included in the skill bundle. That is low-risk in terms of install mechanism, but users should verify the provenance of the xtquant package on PyPI or the chosen source.
Credentials: noteThe skill requests no environment variables and requires only python3. However, it expects access to the local miniQMT userdata path (userdata_mini) and to the miniQMT service; these local files may contain credentials/tokens or broker configuration. Access to those local files is proportionate to the stated trading purpose, but it is sensitive and should be granted only if you trust the skill and the xtquant package.
Persistence & Privilege: okThe skill is not marked always:true and does not request system-wide persistence or modification of other skills. It uses normal autonomous invocation defaults but does not request elevated or permanent privileges.