Platform Corpus
The complete public specification set behind the Subvocal SDK: positioning, architecture, benchmarks, security, and the MCP intent-profile proposal.
The middleware thesis: why software rails, not proprietary hardware.
The v2 config tree, sessions, signal monitoring, telemetry, and auth grants.
The framing shift and what is shipping.
Anatomical, engineering, and systems-level design of the platform.
Shorthand-to-intent reconstruction: providers, prompts, corrections.
The open 50-case shorthand-to-intent evaluation specification.
Latest generated benchmark results across providers and models.
The hardware abstraction layer: replay, synthetic, OpenBCI, Delsys, datasets.
Reference models, training pipeline, calibration, export, quantization.
Authorization policies, threat model, data residency, biometric compliance.
Proposed Model Context Protocol profile for low-bandwidth intent inputs.
Semantic versioning policy, cadence, and changelog format.
Minimum and full bill-of-materials for a DIY sEMG capture rig.
The original feasibility study that seeded the SDK.