tenra Facet

tenra Facet is a search and reframing workbench for turning open-ended questions into inspectable research paths. It emphasizes framing, source boundaries, and answer review rather than presenting search results or AI output as final truth.

Facet is early-stage product infrastructure for research workflows inside the tenra ecosystem.

Operational Purpose

• Help operators reframe a question into more useful research angles.
• Keep search providers, reframing logic, and validation separated.
• Support answer review with explicit assumptions and source boundaries.
• Provide a shared vocabulary for research-oriented handoffs.

Design Posture

• Framing before answering.
• Search provider boundaries over ad hoc scraping.
• Optional geo/realtime packages kept separate until needed.
• Desktop and web surfaces stay thin around shared domain logic.
• AI assistance should improve review clarity, not replace review.

Architecture

apps/
  webapp/       Next.js research and reframing surface
  desktopapp/   Tauri + React/Vite local workbench
  mobileapp/    Expo scaffold for later lightweight review

packages/
  domain/           Question, source, and research-path models
  reframing/        Reframing helpers and answer-shape vocabulary
  search-providers/ Provider interfaces and mock/search adapters
  safety/           Review and boundary helpers
  validation/       Runtime schemas
  api-contracts/    Request and response contracts
  ui/               Shared interface primitives
  config/           Product identity and environment helpers

Current State

• The repo is an active scaffold for the Facet product direction.
• Web and desktop surfaces are present for research workflow development.
• Shared packages define the early product vocabulary.
• Provider behavior is intentionally bounded and still evolving.
• Mobile is scaffolded for later review use.

Deployment Posture

Facet is not production research infrastructure yet. Deployment should wait until provider behavior, source handling, safety boundaries, and persistence expectations are clearer.

Working Locally

pnpm run bootstrap
pnpm run dev:web
pnpm run dev:desktop
pnpm run typecheck
pnpm run verify:all
pnpm run doctor

Direction

• Improve source-boundary and assumption review.
• Clarify which research workflows belong in Facet versus Derive or Sentinel.
• Keep provider adapters replaceable and explicit.
• Build toward structured research handoffs instead of generic answer generation.

Related Documentation

• Architecture Notes
• Negative Spec
• Developer Guide
• Repo Map