Security audit
Security checks for vulnerabilities and agentic risk
The skill is coherent, but it can collect precise location data and send it to external services without clear consent or narrow routing.
Review this skill before installing. It should ask before using current-location sources, prefer manual city or coarse location for weather, disclose which external provider receives location or time queries, and avoid running all services unless the user explicitly requests that.
prompts/01-implement-method.md:16Implicit “All Services” Routing Can Access and Disclose Location Without Explicit User Intent
The description advertises GPS/IP/WiFi-based geolocation and weather/time aggregation but does not warn users that the skill may collect sensitive location data or contact external services. This lack of transparency can lead users to provide or expose precise location information without informed consent, especially when combined with broad triggers.
The trigger list includes generic phrases such as 'location', 'weather', and 'time', which are common in ordinary conversation and can cause the skill to activate when the user did not intend to invoke it. Because this skill can route to location-aware services and potentially external providers, accidental invocation increases the chance of unintended data access or disclosure.
The prompt explicitly directs use of sensitive location sources such as GPS, WiFi, cellular, and IP, but provides no requirement to obtain user consent, minimize collection, or warn that precise location data may be accessed. In a location-aware aggregation skill, this can cause unnecessary collection or disclosure of sensitive geolocation information, especially if the user only asked a vague question.
The instruction to 'execute all services' when intent is multiple or unclear creates an over-collection behavior: it may invoke location, weather, and time services even when the user did not clearly request them. Because location and some time/weather sources may involve network calls and sensitive geolocation lookups, ambiguity is turned into broad data access rather than clarification, increasing privacy and unnecessary external exposure risk.
The documentation explicitly describes collection of precise and coarse location data from multiple sensitive sources, including GPS, WiFi, Cellular, IP, and system signals, but gives no privacy warning, consent guidance, retention limits, or indication that this data is sensitive. In a location-aware skill, this omission can lead operators or downstream users to collect highly sensitive geolocation data without understanding the privacy and surveillance implications.
The workflow explicitly attempts GPS, system, and IP geolocation in sequence, which can collect sensitive location data from multiple sources without any stated user notice, consent gate, or minimization. In a location-aware skill this behavior is functionally relevant, but the absence of disclosure and permission handling creates a real privacy risk because precise coordinates may be gathered unexpectedly.
The weather step sends coordinates to third-party services such as wttr.in and open-meteo, but the workflow does not mention informing the user that their location will be transmitted externally. Sharing location with external providers can expose sensitive movement or residence data and creates a privacy/data-sharing risk beyond local processing.
The specification describes collecting precise location data and configuring optional third-party geolocation providers, but it provides no privacy notice, consent requirement, retention guidance, or warning that location/IP/WiFi/cellular data may be transmitted off-device. Because location data is highly sensitive and the skill is explicitly designed as a unified location-aware service, this omission increases the risk of undisclosed tracking, regulatory noncompliance, and accidental exfiltration to external APIs.
The trigger list includes short, generic terms like "weather," "GPS," and "forecast," which can cause the skill to activate from ordinary conversation without clear user intent. In a location-aware skill, unintended activation is more sensitive because it may lead to external lookups or location-resolution attempts the user did not explicitly mean to invoke.
The intent mapping uses broad tokens such as "where," "when," and "time," so common language can be routed into service execution even when the user is not requesting location, weather, or time data. Because the fallback behavior includes external and potentially sensitive sources, ambiguous routing increases the chance of unintended data access or outbound requests.
The documented fallback chains call out IP, WiFi, Cellular, NTP, and Web API sources, but the skill description does not warn users that external lookups and sensitive location inference methods may be used. This creates a privacy and transparency issue: users may unknowingly trigger network requests or location approximation through less obvious sources when primary methods fail.
The specification describes automatic geolocation using GPS, system, IP, WiFi, and cellular sources plus external weather API queries, but it does not warn users that location and related network data may be collected, inferred, or sent to third-party services. This can lead to privacy harm through unexpected disclosure of precise or approximate location, especially because the workflow includes fallback mechanisms that may still derive location when explicit location is unavailable.
The documentation names external services such as wttr.in, Open-Meteo, NTP, and web APIs, but does not warn that using the skill may send request metadata and possibly user-derived context such as IP-based location or time queries to third parties. While lower severity than direct sensor collection, this can still create unexpected data disclosure and trust issues for users and deployers.
The NTP precision workflow performs an outbound query to pool.ntp.org without warning the user about network access. While this is less sensitive than geolocation sharing, undisclosed external connections can still surprise users, leak metadata such as IP address and timing, and violate restrictive execution or privacy expectations.
No suspicious patterns detected.