Module reference

Every screen in Cloud Discovery, in the order they appear in the sidebar: the core workflow pages, the three analysis tools, and the Library pages.

Dashboard

Your starting point. Three counters (total projects, discovered resources, docs generated) and a list of recent projects with their status. Click any project to jump to its overview.

Projects

All projects as cards showing status, client, project type, resource count, and last update. Two actions per card:

• Duplicate clones a project, useful for what-if scenarios or reusing a structure for a similar engagement.
• Delete removes the project and all of its data after a confirmation dialog. This cannot be undone.

Opening a project shows its overview: the five workflow steps (Upload Data, View Resources, AI Summary, Documentation, Export), each marked Done as you complete it.

Upload Data

Three tabs for getting resource data into a project:

• File Upload: CSV, JSON, or Excel (.xlsx/.xls), one file at a time, up to 10 MB. A new upload replaces existing project data only after it parses successfully with at least one resource.
• Azure Connection: live discovery through a service principal (Tenant ID, Client ID, Client Secret) with a Test Connection check. Quick Scan lists inventory; Deep Scan also inspects configuration for common types (VMs, disks, VNets, NSGs, network interfaces, public IPs, route tables, load balancers, App Service, storage accounts, Key Vaults, and SQL). Re-discovery asks before clearing existing project data.
• Manual Entry: add single resources by name and type to fill gaps.

Resources

The discovered inventory. Counters for resources, resource groups, relationships, and findings; a distribution chart by type; and the resource table (name, type, resource group, location, SKU). Selecting a resource opens a detail panel with its basic properties, deep-scan inspection status and configuration highlights, relationships, findings, and tags.

The Map Relationships button runs the relationship mapper and the findings engine. Relationships come from deep-scan evidence (highest confidence) and rule-based inference; each is labeled with a confidence level. Findings are deterministic observations in four categories (security, reliability, operations, architecture) with severities from high to info, for example missing NSGs, public storage access, unattached disks, or no backup strategy. Re-running the mapping rebuilds both.

AI Summary

One AI call that produces an environment overview split into sections: architecture overview, resource distribution, architectural patterns, risks and concerns, missing documentation, and onboarding notes. Regenerating replaces the whole summary. Requires a license and an AI key.

Documentation

Generates the report content, section by section, from one of four templates: Technical Operations Report (12 sections), Architecture Handoff Report (9), Executive Summary (5), or Migration Assessment Report (9). A plain 11-section default template is also available. Each section is saved as soon as it is generated, so an interrupted run keeps its completed sections.

• Sections are Markdown. Toggle between preview and an editor per section; saved edits are marked with an edited badge.
• Regenerate a single section, or use Regenerate All to start over with a different template, for example an executive version of the same environment.

Export

Turns the generated documentation into a deliverable. The content comes from the Documentation page; the Export page controls how it looks:

• Visual styles: a gallery of 14 styles grouped into Modern, Classic, Technical, and Expressive, from Swiss Minimal and Executive Classic to Technical Mono and KPI Dashboard.
• Options: cover page, findings section, resource appendix, and compact mode toggles, plus customization for organization name, footer, accent color, font style, and logo.
• Preview: renders the actual PDF bytes, so the preview is identical to the exported file.
• Formats: Professional PDF, a basic Simple PDF, or Markdown with a table of contents. Files save through the native macOS dialog and are recorded on the Reports page.

Scout, the AI Analyst

Scout is a per-project chat that answers questions about the loaded environment. It is grounded: instead of answering from general knowledge, it investigates your actual data through four read-only tools (list resource types, list resources, get resource details, get findings) and cites what it found. For security questions it always pulls the deterministic findings first, the same findings that appear in your exported report.

• What it answers: anything about this project's environment: inventory, architecture, dependencies, configuration, security posture, migration readiness. It can also draft work products from that data, such as client emails, tables, or talking points.
• What it declines: anything else. General programming help, unrelated facts, creative writing, and personal advice are out of scope by design; Scout declines in one sentence and offers to help with the environment instead.
• Transparency: while Scout investigates, a chip shows each tool call as it happens. Every answer keeps a collapsible "Investigated" log listing each lookup and a preview of what came back, so you can verify the answer against the evidence.
• Model picker: the model pill under the composer switches models for the whole app, not just the chat.

If an answer is cut off, ask a narrower question; Scout has a fixed investigation budget per question.

Topology Draft

An editable network diagram generated from your resources and relationships. Resources are nested inside resource group containers, laid out automatically, with Azure icons and a confidence marker (confirmed, inferred, ambiguous) on each node.

• Editing: drag nodes, add resources from the inventory panel, delete with the Delete key, and adjust a node's confidence in the details panel. Auto Layout and Zoom to Fit are in the toolbar.
• Topology Notes: the right panel lists warnings (for example, no NSG found) and the assumptions the generator made, each dismissible.
• AI Architect: with a license and AI key, this mode designs a presentation-ready view: it picks the 8 to 15 meaningful nodes, groups them into tiers, hides plumbing such as NICs and route tables, and infers connections. The result is always marked as an inferred draft for you to review and correct.
• Export: PNG (high resolution), PDF (landscape, labeled as a draft), or JSON of the diagram data. These diagram exports are not recorded on the Reports page.
• Saving: Save changes persists the diagram locally so it is restored the next time you open the project. The AI Architect saves its draft automatically.

Migration Assessment

A rules engine that classifies whether each resource can move, based on Microsoft's documented move support. Choose a scenario, Move to Different Subscription or Move to Different Resource Group (same sub), optionally narrow the scope to one resource group, and run the assessment. No AI key needed.

• Classification: each resource lands in one of six categories: Supported, With Prerequisites, Conditional, Not Supported, Recreate, or Review Needed, with notes, prerequisites, and limitations from the rules.
• Dependency groups: resources linked by strong relationships are grouped as "move together" sets.
• Blockers and complexity: a blockers card lists what cannot move, and the whole assessment gets a complexity rating from Low to Very High.
• Exports: the classification table exports to CSV. With a license and AI key, AI Migration Analysis writes a narrative readiness report you can export as PDF; it is saved with the project so it survives navigation.

The rules behind the assessment are browsable on the Move Rules page.

Library

Reports

Every deliverable exported from any project, in one place: file name, project, client, format (PDF, Markdown, CSV), and export date, with search across all of them. Per row you can copy the full file path or reveal the file in Finder. The list is a history of where files were saved; if you move a file afterwards, the Finder shortcut will no longer find it.

AI Usage

Local accounting of every AI call: total tokens, call counts, and prompt versus completion tokens for a selected month, broken down by project, by operation (summary, documentation, migration analysis, AI Analyst, topology architect), and by provider and model. The app shows tokens, not dollars; exact billing lives in your provider's dashboard. Nothing on this page is sent anywhere; it is recorded on your machine.

Move Rules

The knowledge base behind the migration assessment: 752 Azure resource types, of which 34 are curated in depth (with prerequisites, limitations, and dependency notes) and 718 come from Microsoft's move-support matrix. Search by resource type or name, filter by classification or tier, and open the Microsoft Learn source for any rule.

Settings

License activation, AI provider and model, API key management, a token usage snapshot for the current month, and demo data. Covered in detail in installation and license and AI setup.