What Luckline does

Luckline helps you find the right person already connected to your network. Enter a natural-language search, and it ranks matching people using role fit, source overlap, group membership, and relationship warmth. Each result shows its supporting signals, source provenance, and the warmest introduction path. You can inspect a research profile, draft an introduction, copy a brief, or export results to CSV.

The included demo uses clearly synthetic professional-network data. Live third-party connectors are intentionally disabled.

What changed during Build Week

Luckline existed as a June prototype before the event. During Build Week, I used Codex with GPT-5.6 to add the local-only private pilot represented by commit 92468aa. This is the work being submitted for evaluation:

  • One-command private startup with npm run pilot
  • A trusted Node.js signer/gateway bound only to 127.0.0.1
  • Fixed local user and tenant identity with fresh timestamped HMAC signatures
  • Forged inbound identity headers stripped and overwritten
  • HMAC secrets kept in server memory and never exposed to browser code
  • Tenant-scoped private storage initialized empty with mode 0600
  • CSRF protection, consent filtering, stale-signature rejection, and truthful connector-unavailable behavior
  • End-to-end private-mode support for search, profiles, intro drafting, and CSV export

The pre-existing search interface provides the product context; the secure local pilot is the Build Week extension.

How I used Codex and GPT-5.6

GPT-5.6 in Codex helped trace the existing browser-to-server authentication flow, identify the smallest safe insertion point, drive TDD from a failing forged-header test to the completed gateway, and perform adversarial security and complexity reviews. It also exercised the real browser flow and verified the private-session, search, export, restart, and unavailable-sync states.

Run it

Use Node.js 18 or newer:

npm run pilot

Then open http://127.0.0.1:4173. The pilot starts with an intentionally empty private dataset. For the populated synthetic demonstration, run npm start.

Built With

Share this project:

Updates