geospatial-telematics-developer
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill package is composed entirely of instructional documentation. No scripts, binaries, or active execution components are present within the files.
- [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for ingesting untrusted telemetry data from external sensors and fleet devices. While this creates a potential surface for data-level prompt influence, the skill provides specific guidance on validation and filtering to mitigate these risks.
- Ingestion points: External telematics feeds via MQTT and Kafka as described in
references/telematics_ingestion_and_quality.md. - Boundary markers: The instructions mandate normalization into canonical schemas and validation against device specifications.
- Capability inventory: The skill provides logic templates for spatial SQL and API construction; it does not contain tools for direct command or code execution.
- Sanitization: Defines strict quality dimensions (accuracy thresholds, teleport detection, jitter filtering) to sanitize incoming data streams.
Audit Metadata