A clear local-first boundary for Salesforce assessment data.
SF Analyzer separates customer org evidence from product operations: the desktop app stores Salesforce data locally, while the portal handles account and license workflows.
Data Flow Diagram
The diagram is intentionally simple: customer evidence moves from Salesforce to the local desktop workspace; AI calls go directly to the provider selected by the user.
Salesforce org → desktop app
The desktop app authenticates with the local Salesforce CLI and retrieves metadata/source files to the user's machine.
Desktop app → local workspace
Snapshots, source files, git-backed history, wiki pages and exported documents are written under the local ~/.sf-analyzer workspace.
Desktop app → AI provider
Only when AI generation is run, selected metadata/code context is sent directly to the provider configured by the user with their own key.
Desktop app ↔ portal
The portal handles account, license, limits and operational telemetry. It is not the storage layer for Salesforce org data.
Data Boundaries
The portal is not a repository for customer org metadata or source code. AI provider traffic is direct BYO-key traffic controlled by the user.
- •Salesforce metadata snapshots and retrieved source files
- •Local git analysis history and delta evidence
- •Generated wiki Markdown pages and exported technical documents
- •LLM API keys stored locally with OS-backed encryption
- •Account identity
- •License status and usage limits
- •Operational telemetry and bug reports when enabled
- •Prompt instructions for As-Is documentation
- •Selected local metadata/code excerpts needed for the requested AI output
- •No SF Analyzer cloud proxy for customer prompts or source code
Application Hardening
The desktop app follows Electron hardening practices documented in ADR-007: sandboxed renderer, typed IPC and OS-backed secret storage.