Security audit
Security checks for vulnerabilities and agentic risk
This is a coherent NL2SQL reporting workflow that performs bounded read-only database querying, so users should enable it only where agent-driven database reads are acceptable.
Install this only for agents connected to databases where read-only querying, schema inspection, and small supplemental analysis queries are acceptable. Use database-side read-only roles and avoid connecting highly sensitive tables unless users expect the agent to analyze them.
The manifest description and trigger text are entirely in Chinese and frame the skill behavior in that locale, with no indication that users may choose another language. Under the policy, a forced language or locale without opt-in is a natural-language policy concern unless the locale restriction is explicitly justified.
The trigger conditions are broad enough to match many ordinary data-assistance requests, so the skill may activate when a user did not explicitly intend to run an NL2SQL workflow. In this context, over-broad activation is risky because later stages execute SQL and may issue additional exploratory queries, increasing the chance of unintended access to connected data sources.
The workflow explicitly includes SQL execution, automatic retries, ReAct-based database exploration, and even 1-2 supplemental queries for reporting, but the description does not warn users that connected data sources will be queried. This lack of transparency can lead to unintended database activity, surprise data exposure, and excessive querying beyond the user's expected scope.
No suspicious patterns detected.
