Export, query, and explore your SignalK data archived as Parquet files
Manage which SignalK paths to collect data from:
Register and manage SignalK commands for remote control:
Analyze your maritime data using Claude AI for insights, patterns, and recommendations.
Click to expand - context automatically included with every AI analysis
Choose the Claude model for analysis. Sonnet provides the most detailed insights.
View and manage your previous Claude AI analyses.
Validate parquet file schemas against SignalK metadata and schema guidelines:
Test your cloud storage connection (Amazon S3 or Cloudflare R2) and sync files:
Key Prefix organizes your uploaded files in the bucket with a directory structure:
tier=raw/context=vessels__urn.../path=navigation__position/year=2025/day=193/file.parquet
marine-data/tier=raw/context=vessels__urn.../path=navigation__position/...
boat-123/tier=raw/context=vessels__urn.../path=navigation__position/...
This allows organizing multiple vessels or data sources in the same bucket. Works identically for S3 and R2.
Compare local files with cloud storage and sync missing files.
Available SignalK data paths with Parquet files:
Run a query to see results here...
Spatial querying and visualization of historical SignalK data. Draw a search area, configure your query, and explore results on the map.
Parquet store overview, SQLite buffer status, and data migration tools.
Loading store stats...
Loading buffer status...
Convert existing flat-structure Parquet files to Hive-style partitioned structure for better query performance.
Import historical GPS tracks from .gpx files into the parquet store. Each track point is expanded into SignalK records for the selected paths and written as raw-tier parquet files.
Merges per-day parquet files within each (tier, context, path, year) group into a single per-year file, then deletes the source day-files. Run this on a long-running store to keep DuckDB queries fast and the filesystem light. Only completed years (strictly before the cutoff) are touched.