Extension Roadmap¶
Target Architecture — Final-State Design
This page describes how the final-state Factory Studio is designed to grow. The microfrontend shell, MFE contract, and BFF tier that make this extensibility possible are real today (ConnectSoft.Blazor.ShellTemplate, ConnectSoft.Blazor.MicrofrontendLibraryTemplate, ConnectSoft.Blazor.UIKit); the modules, services, and APIs below are the planned extension surface.
The Factory Studio is built to grow without a monolith. Because every module is an independently versioned Blazor microfrontend over a thin BFF, and because the Studio owns no source-of-truth data, new capabilities are added by composing new MFEs and BFFs onto the existing shell rather than by rewriting it. This page defines the principles for that growth and the concrete extension opportunities.
Extension principles¶
- Microfrontend-first. New experiences ship as new MFEs built on
ConnectSoft.Blazor.MicrofrontendLibraryTemplateand composed by the shell. The shell,Blazor.UIKit, OpenIddict session, and SignalR client are shared; modules are added without a monolithic redeploy. - BFF-per-experience. A new module that needs aggregation gets its own
ConnectSoft.Factory.Studio.*BFF (or a new endpoint on an existing one), keeping each BFF cohesive and independently scalable. The Studio never reaches a source-of-truth store directly. - Aggregation, never ownership. Extensions read from source platforms and write back only governed human-authority commands. New modules must not become a second system of record.
- Governed by default. Every new module, endpoint, and stream inherits tenant isolation, RBAC, and audit; an extension is not "done" until it has a role mapping and emits audited events.
- Traceable and observable. New surfaces carry
traceId/correlationIdthrough every read and command and emit canonical-envelope events, so the extended experience stays correlatable end to end. - Reuse over rebuild. New modules should compose existing BFF read models and
Blazor.UIKitcomponents before introducing new ones, extending the factory's reusability principle to the experience layer. - Marketplace-extensible. Third parties and tenants can contribute experience extensions (modules, widgets, view packs) through the Marketplace, subject to governance review.
Future UI modules¶
| Module | Purpose | Primary sources | Notes |
|---|---|---|---|
| Intent Studio | Capture and refine business intent into blueprints conversationally and visually. | Control Plane, Knowledge Platform | Feeds the Blueprint Designer; not a chatbot — a structured intent surface. |
| Experiment Center | Run and compare A/B prompt, model, and pattern experiments on factory output. | Observability & Feedback, Agent Mesh | Surfaces self-improvement loop results. |
| Compliance Evidence Room | Assemble audit-ready evidence packs per framework (SOC 2, ISO, etc.). | Governance, Security & Compliance | Built on the immutable decision + audit trail. |
| Reuse Marketplace Browser | Discover reusable patterns, modules, and blueprints across the tenant estate. | Knowledge Platform, Template & Library | Differs from Install Center: discovery of internal reuse. |
| Stakeholder Portal | Read-only, executive-friendly portfolio and outcome views. | Dashboard aggregation | Curated Viewer-role experience. |
| Incident War Room | Live, collaborative incident response over runtime signals. | Observability & Feedback, Runtime Cloud | Real-time multi-user session over SignalR. |
| Mobile Companion | On-the-go approvals and alerts. | StudioBff | Built on ConnectSoft.MauiBaseTemplate; focuses on review + notifications. |
Future BFF services¶
| Service | Responsibility |
|---|---|
ConnectSoft.Factory.Studio.ExperimentBff |
Aggregate experiment definitions, runs, and comparative results for the Experiment Center. |
ConnectSoft.Factory.Studio.ComplianceEvidenceService |
Compose audit evidence packs from governance, review, and lineage data. |
ConnectSoft.Factory.Studio.ReuseDiscoveryBff |
Surface cross-project reuse candidates from the Knowledge Platform and Template & Library platform. |
ConnectSoft.Factory.Studio.CollaborationGateway |
Multi-user collaborative editing/presence (war room, shared reviews) over SignalR. |
ConnectSoft.Factory.Studio.IntentBff |
Aggregate intent capture, refinement, and blueprint candidate generation. |
ConnectSoft.Factory.Studio.ExperienceExtensionHost |
Sandbox and serve marketplace-contributed experience extensions safely. |
Future APIs¶
- Studio Extension API — a stable contract for registering marketplace-contributed microfrontends, widgets, and saved-view packs, with capability and permission declarations.
- Saved View Pack API — import/export of curated
SavedViewandDashboardViewbundles between tenants and from the marketplace. - Webhook / outbound event API — let tenants subscribe external systems to Studio events (review outcomes, escalations) for integration with their own tooling.
- Embedding API — signed, scoped embeds of individual Studio modules (e.g. a runtime health board) into external portals.
- Collaboration presence API — presence, cursors, and shared selection for multi-user sessions.
Marketplace opportunities¶
- Experience packs — installable module bundles (e.g. an industry-specific compliance evidence room) distributed through the Marketplace.
- Widget catalog — third-party dashboard widgets that bind to allowed BFF read models, governed by the Extension API permission model.
- View pack library — shareable
SavedView/DashboardViewconfigurations for common roles and industries. - Theme & branding packs — tenant white-labelling of the shell and
Blazor.UIKitfor partner-facing portals. - Connector-driven modules — modules that surface data from Integration Platform connectors alongside native factory data.
Agent opportunities¶
The Studio is a natural surface for agents that assist the human in the loop — always under the same RBAC, tenant, and audit guarantees, and never bypassing human authority:
- Review assistant agent — pre-analyzes a
ReviewRequest, summarizes the artifact diff and lineage, flags risks, and proposes a recommendation the human confirms, accelerating (never replacing) the decision. - Dashboard insight agent — watches the live event stream and surfaces anomalies and "what changed and why" narratives on the command center.
- Escalation triage agent — clusters and prioritizes the review queue by risk and SLA, routing escalations to the right
Approver. - Knowledge navigation agent — turns a natural-language question into a bounded knowledge-graph exploration or artifact lineage walk, returning an access-checked view.
- Cost optimization agent — proposes concrete optimizations in the Cost Center from FinOps signals, each as a reviewable recommendation.
- Onboarding agent — guides new operators through the Studio, configuring their workspace and saved views per role.
Every Studio-surfaced agent operates through the same BFF and command path as a human, emits audited envelope events, and respects the security model — agents propose, humans retain authority.