Back to workspace

Product and API guide

Mnemosyne, explained clearly

Mnemosyne helps you compare source artifacts, inspect evidence-backed resonances, and turn a selected idea into a runnable prototype.

1. Add a corpus

Upload code, text, documents, images, audio, or archives. Mnemosyne keeps the workspace local-first and shows parsing limitations when they apply.

2. Run analysis

Choose a reasoning mode and provider. Every run records its provider, model, evidence coverage, and any graceful fallback.

3. Review and synthesize

Inspect evidence, tensions, and next questions before generating or opening a prototype from a resonance.

Provider behavior

  • Smart route selects a configured provider, then uses the deterministic local engine when no provider is configured or a provider is unavailable.
  • OpenAI and Google Gemini are optional server-side integrations. Browser code never receives their credentials.
  • Local engine is deterministic and available without an external provider, making the demo reviewable even during provider capacity events.

Application endpoints

EndpointPurpose
POST /api/ingestParses supported source files into corpus artifacts.
POST /api/analyzeBuilds an evidence-aware resonance analysis with provider traceability.
POST /api/generate-prototypeGenerates a runnable prototype or preserves the deterministic local version.