Dashboard/Scout (Second-Brain)/wiki/decisions/2026-06-29-private-beta-launch-surface.md
Scout (Second-Brain)

wiki/decisions/2026-06-29-private-beta-launch-surface.md

ADR: Scout Private-Beta Launch Surface

Date: 2026-06-29 Status: Accepted

Context

Scout now has a broad implementation surface: local package, CLI, FastAPI service, Docker service, hosted beta API, Claude/Codex skill backend, product export tools, launch website, documentation, and private-beta handoff material. The risk is over-claiming public readiness while public-launch blockers remain open.

Decision

Scout's launch surface is controlled private beta with explicit limits, not public launch.

Rationale

  • Verified evidence supports local/core usage, Docker smoke, hosted beta quickstart, website rendering, product exports, skill packaging, and beta onboarding.
  • The release checklist still blocks public launch on license approval, final license expression, public pricing, registry publishing policy, Crawl4AI/lxml risk acceptance or mitigation, real Stripe test-mode smoke, and final terms/privacy.
  • The legacy /app UI is not a launch surface. The launch website and HTTP/CLI/skill surfaces are the current product-facing surfaces.
  • Hosted Scout should be finite-credit and metered; do not sell unlimited hosted crawling.

Alternatives Considered

Option Why rejected
Public launch now Legal, dependency, payment, registry, and public hosted operations gates remain open.
Launch only the website The website is useful only if it points to real install/API/beta flows and truthful limits.
Treat the legacy app UI as the product The app UI was not validated as an end-to-end product surface and should not be sold or presented as ready.

Consequences

  • Private beta testers can use local install, Docker, hosted beta API, skill usage, and website docs.
  • Public launch claims must cite docs/product/launch-evidence-index-2026-06-29.md.
  • Founder decisions must close before registry publishing, broad hosted access, or public commercial launch.
  • PRISM and CI should consume Scout as acquisition infrastructure, not as a finished intelligence interpretation product.