Skip to content

jmars319/Assembly

Repository files navigation

tenra Assembly

tenra Assembly is an AI-assisted content operations workbench for turning source material into reviewed, human-approved content outputs. It keeps draft generation, review state, approval, export, and provider configuration visible instead of treating content production as a black box.

The system is desktop-first for local workbench use, while the existing web app remains the fuller cloud-backed implementation for auth, Prisma-backed workflows, and hosted API work.

Operational Purpose

  • Convert briefs, notes, and source material into structured draft workflows.
  • Keep human review and approval as explicit workflow states.
  • Support local drafting and export without requiring a cloud provider.
  • Share reusable domain, prompt, and type logic between web and desktop surfaces.

Design Posture

  • AI assistance is a bounded drafting aid, not an autonomous publisher.
  • Local model support is available through provider boundaries.
  • The desktop surface owns portable workbench import, export, and review loops.
  • Hosted-web concerns stay isolated in the web app.
  • Shared packages carry product vocabulary and validation rules.

Architecture

apps/
  webapp/       Next.js, Prisma, auth, API routes, and fuller product behavior
  desktopapp/   Tauri + Rust + React/Vite local content workbench

packages/
  shared-types/ Reusable enums, DTOs, presets, and feature keys
  domain/       Validation helpers, parsers, and audit-label mapping
  prompts/      Shared instruction layering and prompt guidance

docs/           System overview, desktop architecture, repo map, and handoffs

Current State

  • The web app remains the broadest product surface.
  • The desktop app is a usable local workbench for manual content items, review gates, approvals, Markdown export, and JSON workbench backup/restore.
  • Local AI provider support is available through Ollama or OpenAI-compatible endpoints.
  • OpenAI-compatible hosted provider support remains available when configured.
  • The desktop app does not bundle model weights.

Deployment Posture

Assembly is currently a mixed local and web codebase. The desktop app is suitable for local operator workflows; the web app owns hosted implementation concerns. Production hardening depends on the chosen deployment target, provider configuration, auth posture, and database setup.

Working Locally

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

For local model drafting, configure the ASSEMBLY_LOCAL_AI_* environment values described in the developer guide.

Direction

  • Continue separating local workbench ownership from hosted-web responsibilities.
  • Improve source tracking, review ergonomics, and export formats.
  • Keep provider configuration explicit and replaceable.
  • Preserve human approval as a required operational boundary.

Related Documentation

About

Desktop-first content operations workbench for AI-assisted drafting, review, approval, and export.

Topics

Resources

Stars

Watchers

Forks

Packages

 
 
 

Contributors