Calm, tenant-aware clients
Web and sync experiences keep navigation immediate while every workspace request carries explicit tenant context.
System design
Cinch is a working proof of concept for a resilient, multi-tenant collaboration system—and for the disciplined operating model used to build it.
System map
Each layer has one job, a visible trust boundary, and evidence that can fail independently.
Web and sync experiences keep navigation immediate while every workspace request carries explicit tenant context.
Account discovery, workspace traffic, and synchronization are separate surfaces with explicit contracts and independent health signals.
Tenant identity is enforced through the read, write, and sync paths; durable records preserve the context behind team decisions.
Infrastructure as code, release gates, external checks, and recovery drills turn operating assumptions into reviewable artifacts.
Resilience model
Resilience is the ability to prevent avoidable failures, notice real degradation, recover deliberately, and retain the lesson.
Host isolation, scoped sync tokens, least-privilege deployment identities, and repository boundaries reduce the blast radius before failure.
External checks cover the public root, demo journey, sync path, and operations surface—not just whether a process is running.
Restore, rollback, and replica resynchronization paths are treated as executable procedures and rehearsed independently.
Incidents and meaningful tradeoffs feed tests, operating notes, and ADRs so the system retains what the team learned.
Deployment strategy
The target operating model lets the operator approve a defined outcome, then allows an agent to build and deploy independently inside that authorization.
Agent turns demand into scope, risks, acceptance criteria, and an ADR when the choice is consequential.
The operator approves the delivery envelope and release authority before implementation begins.
Agent implements within a bounded repository, preserving system and trust boundaries.
Automated gates assemble build, test, security, journey, and recovery evidence.
The exact validated revision is promoted as a versioned deployment with rollback retained.
Approved architecture evidence
Only operator-reviewed application claims appear here, with their exact maturity and remaining limits intact.
Application releases use a blue/green strategy. The complete replacement web fleet and fresh SBOM-backed image policy must pass before traffic is promoted, the prior image is retained as rollback evidence, and routine application releases leave the real-time sync service unchanged.
Health-gated promotion reduces release risk for authenticated customers while separating routine product delivery from maintenance on the stateful real-time sync tier.
Releases on 2026-08-10 exercised healthy replacement Machines, immutable identity, public checks, the private Zero hop, and unchanged Zero. On 2026-08-11 refreshed vulnerability data blocked two candidates without an exception or promotion. Later, evidence-backed OpenSSL dispositions yielded zero blocked findings; source 70e27f0ffa484fc0964f0672fc08a567d058adb0 passed the same gate, promoted two healthy Machines, passed all twelve public checks, and left Zero unchanged. Active-session continuity remains an evidence gap, so zero-interruption wording is not approved.
Infrastructure components
Cookie-free public content with an independent build and rollback history.
Central sign-in and workspace discovery with host-only sessions.
Workspace-local sessions and explicit tenant routing.
Short-lived, tenant-scoped access for real-time synchronization.
Isolated identities, preview releases, immutable revisions, and controlled promotion.
Boundary tests, security checks, health probes, benchmarks, and recovery records.
Code principles
Critical architecture rules belong in tests and deployment policy, not only in diagrams.
Small interfaces, explicit state, idempotent operations, and reversible changes make recovery understandable.
Local-first feedback and optimistic updates are paired with authoritative reconciliation and measurable sync behavior.
ADRs record the context, alternatives, decision, consequences, and evidence for consequential changes.
Agents work within named repositories, scopes, and acceptance criteria; the operator controls intent and release authority.
A passing build is necessary, but production readiness also includes isolation, security, journey, and recovery evidence.
Next: trust model