tenra Scout

tenra Scout is a local-first market scanner for finding, auditing, and classifying business web presences. It turns a narrow market query into a queued acquisition run, deterministic website audit, evidence set, and operator-readable opportunity report.

Scout is not an SEO platform or a general crawler. The product shape is intentionally operational: define the market, run the scan, inspect evidence, and decide what outreach or follow-up is justified.

Operational Purpose

• Identify local or vertical-market businesses with weak or missing web presence.
• Preserve enough evidence for a human operator to understand why each candidate was surfaced.
• Separate acquisition, presence typing, audit, shortlist ranking, and outreach drafting into reviewable steps.
• Keep AI assistance optional and grounded in stored Scout evidence.

Design Posture

• Deterministic audit logic before generative assistance.
• Local desktop workflow as the primary operator surface.
• Postgres-backed run queue and repository layer for repeatable runs.
• Explicit confidence and evidence instead of opaque lead scoring.
• Thin app surfaces with shared domain, validation, privacy, and UI packages.

Architecture

apps/
  webapp/       Next.js runtime, reports, run control, and local web surface
  desktopapp/   Electron shell that runs the local web app and worker
  mobileapp/    Scaffold for later lightweight review

packages/
  domain/       Market, candidate, audit, and shortlist models
  api-contracts/ Request and response contracts
  validation/   Runtime schemas
  privacy/      Redaction and safe-display helpers
  ui/           Shared interface primitives
  config/       Product identity and environment helpers

scripts/        Bootstrap, worker, verification, and packaging flows
docs/           Repo map, developer guide, handoffs, and stability notes

Current State

• The desktop app is the primary product surface.
• Search acquisition uses bounded provider adapters rather than open crawling.
• Runs are queued and stored through Postgres.
• Audits use Playwright and accessibility checks with locally stored screenshot evidence.
• Outreach packs are supported for shortlisted businesses, with deterministic local drafting and optional Ollama or OpenAI assistance.
• The mobile app is present as a future surface and is not the active product.

Deployment Posture

Scout is currently developed as a local desktop and local web workflow. Packaged desktop builds are supported for local macOS use, with ad-hoc signing when Apple signing and notarization credentials are not present. The project is not positioned as a hosted SaaS service.

Working Locally

pnpm run bootstrap
pnpm run db:prepare
pnpm run dev:all
pnpm run dev:desktop
pnpm run verify:all
pnpm run doctor

Use pnpm run worker:start when running the queue worker separately from the managed desktop development flow.

Direction

• Keep acquisition narrow and explainable.
• Improve evidence quality and operator review ergonomics.
• Strengthen persistence, import, and export workflows.
• Keep AI-generated outreach clearly marked as assistance, not automated authority.

Related Documentation

• Developer Guide
• Product Overview
• Repo Map
• Stability Checklist