Cable Intel

Cable Intel helps you identify cables quickly, label them consistently, and avoid trial-and-error.

What It Does

• Identifies cables from catalog data or free-text printed markings.
• Infers connector/power/data/video capabilities and summarizes expected behavior.
• Suggests deterministic physical label colors for strap and holder classes.
• Supports catalog filtering (brand, connector type, length, color, price bucket).
• Includes a TUI for ingest discovery, seeding, and quality checks.

How It Works

• apps/web: SvelteKit workspace UI (catalog + manual flows).
• packages/backend: Convex schema/actions/queries and ingest pipeline.
• packages/shopify-cable-source: Shopify source discovery/extraction.
• apps/tui: terminal ingest manager.
• Preview deployments run preview-validation (seed + runtime QA + browser smoke with artifacts/replay links).

Quick Start

mise install
mise exec -- bun install
bun run dev:setup
bun run dev

Web app: http://localhost:5173. mise.toml pins Node 24.21.0 and Bun 1.4.2 for local work and CI; Vercel uses an explicit bunx bun@1.4.2 install/build command and the Node 24 engine. Its build script verifies the actual Bun version against packageManager.

Testing And CI

Layer	Present	Tooling	Runs in CI
unit	yes	bun run test:unit	yes
integration	yes	bun test integration suites	no
e2e api	yes	bun test packages/backend/convex/ingest.e2e.test.ts	no
e2e web	yes	pinned agent-browser smoke scripts	local fields in CI; inference/reset on preview

bun run check-types checks the web, backend, source package, and TUI. For local checks without a backend, set PUBLIC_CONVEX_URL=https://placeholder.convex.cloud. The local browser check deliberately disables Convex and does not claim inference or reset coverage. Live source/provider integration tests remain separate and need network access and the documented provider credentials.

Primary CI workflows:

• .github/workflows/ci.yml (quality + local browser smoke + build)
• .github/workflows/preview-seed.yml (preview seed + browser QA)
• .github/workflows/org-required-checks.yml (required merge gate)

Deployment And External Services

• Deploy target: Vercel (vercel.json)
• Data/backend: Convex
• LLM path: AI Gateway/OpenAI-compatible via ai SDK
• Optional analytics: PUBLIC_VERCEL_ANALYTICS_DSN (already wired in apps/web/src/routes/+layout.ts)

Preview automation secrets:

• CONVEX_DEPLOY_KEY_PREVIEW
• BROWSERBASE_API_KEY
• BROWSERBASE_PROJECT_ID
• VERCEL_AUTOMATION_BYPASS_SECRET

API Surface

There is no public REST API in this repo. App/backend integration is via Convex actions and queries.

Active Backlog

• Search quality phase 2: semantic rerank + measured eval (#92)

License

MIT License. See LICENSE.

Copyright 2026 SHPIT LLC.