SignalStitch
SignalStitch is a Rust-first desktop application for matching and linking noisy automotive data across module inventories, configuration records, and observed CAN diagnostics traffic.
Core Direction
The matching engine is designed to stay deterministic and explainable, with source lineage preserved so every suggested link can be reviewed, traced, and corrected by a human.
Primary Stack
A Rust workspace for the engine and data model, SQLite persistence for sessions and evidence, and a Tauri desktop UI for review workflows.
Planned Capabilities
- Import and normalize module inventory data, config module data, and parsed CAN traffic in one session
- Generate candidate links with weighted evidence and explicit threshold bands
- Expose why a match was suggested instead of relying on opaque scoring
- Support human review, manual linking, rejection, and audit trails in a desktop workflow
- Resolve accepted links into stitched module identities without losing raw lineage
- Version rules, thresholds, and learned priors so results remain reproducible over time
Why It Matters
SignalStitch is aimed at the messy part of diagnostics and reverse-engineering work: reconciling partial, inconsistent, and overlapping data without turning the matching process into a black box.