gis-mapping
Fail
Audited by Gen Agent Trust Hub on Feb 22, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
- REMOTE_CODE_EXECUTION (HIGH): The skill implements dynamic script loading mechanisms that fetch and execute JavaScript from external sources at runtime. This bypasses static security controls and allows for potential code injection if the external source is compromised.
- Evidence:
references/leaflet-arcgis-equivalents/integration-guide.mdcontains aloadScriptfunction and a list of CDN URLs (unpkg.com) loaded sequentially.references/leaflet-arcgis-equivalents/05-heatmaps-density.mdalso dynamically appends a script tag forleaflet-heat.js. - COMMAND_EXECUTION (HIGH): The PHP code examples for spatial database queries are vulnerable to SQL Injection. Variables are interpolated directly into raw SQL strings instead of using prepared statements.
- Evidence: In
references/leaflet-arcgis-equivalents/02-spatial-queries.md, thefindInBuffermethod interpolates$lng,$lat, and$bufferMetersdirectly into the$bufferWKTstring used in awhereRawcall. - COMMAND_EXECUTION (HIGH): The skill provides patterns for generating JavaScript from PHP that are vulnerable to Cross-Site Scripting (XSS). User-controlled or configuration variables are placed directly into script blocks.
- Evidence:
references/leaflet-arcgis-equivalents/01-basic-mapping.mduses a heredoc ingetMapScriptto generate JS, interpolating{$elementId}and configuration values directly into the script. - DATA_EXFILTRATION (MEDIUM): The skill makes external requests to various third-party GIS providers (Nominatim, Photon, OSRM, ArcGIS) with location data. While expected for GIS functionality, there is no evidence of data sanitization or proxying to protect user privacy.
- Evidence:
03-geocoding.mdand06-routing-network-analysis.mddemonstrate directcurlandfetchcalls to external domains. - CREDENTIALS_UNSAFE (LOW): The skill references the use of API keys (e.g.,
osm_api_key) and demonstrates a pattern of exposing these keys to the client-side via thewindowobject. - Evidence:
SKILL.mdandgeofencing.mdshow patterns for injectingwindow.osmApiKeyfrom system settings.
Recommendations
- AI detected serious security threats
- Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata