guteneo turns exact documents into reviewable communications. A browser, native client or connected assistant can prepare work; the server decides which actor may read it, approve its immutable version and request delivery. PDF verification, provider readiness, budgets and delivery evidence remain separate checks.
This site explains the implementation and the decisions behind it. It serves engineers and integrators; customer task guides belong to the separate future customer documentation site.
Run the application
Start the real Workers application with local D1, private document storage and synthetic data.
Understand the system
Follow the browser, shared domain service, private PDF services, outbox and delivery queues.
Explore the feature atlas
Browse the complete generated implementation inventory, including actors and activation gates.
Build an integration
Use the same document and dispatch contracts through REST or MCP without bypassing approval.
The commitments that shape the code
Preserve the original. Uploaded PDF bytes have a stable hash. Rendering, address-page creation and Studio generation produce distinct document versions; they do not silently modify an approved original.
Keep permission and approval distinct. Workspace membership and OAuth scopes control access. Approval binds exact content, recipient, options and cost. An assistant connection is not human consent; optional delegation is a separate, bounded administrator decision.
Commit before dispatching. Acceptance, quota reservation and outbox insertion are atomic. Queue publication can repeat; a conditional claim controls provider attempts. An unknown provider outcome requires reconciliation rather than blind retry.
Describe the evidence. Implemented code, an enabled account capability, a local fixture, a deployed release and a verified delivery prove different things. Feature pages retain that distinction. They are an engineering inventory, not a promise that every account, host or country is currently enabled.
Choose a reading path
-
New to the repository: start with the domain model, then use the repository map.
-
Tracing stored relationships: read the data model and ownership for tenant boundaries, immutable versions, accounting and retention.
-
Working on document creation: follow PDF handling, Studio and data and PDF diagnostics.
-
Working on delivery: read the dispatch lifecycle before the channel adapters.
-
Adding a client or assistant: understand authority and tenancy, then choose REST or MCP.
-
Operating the platform: read Belvédère administration for its seven views, privileged access and read-only boundaries.
-
Following a flow visually: open the technical diagram reference for the interactive figures, Mermaid flows and downloadable sources.
-
Changing the implementation: use the testing boundaries and contribution workflow.
-
Preparing a release: follow changes and releases from reviewed source through CI to exact served verification.
-
Investigating a failure: use troubleshooting to separate safe reads, explicit recovery and operator escalation.
-
Working on personal receipt: follow secure distribution for independent account association, human approval and personal-copy retention.
-
Maintaining deployment and costs: use GitHub CI and release evidence, provider and tariff maintenance and background jobs.
-
Creating a compliant interface or film: start with the design system, AI contribution rules and video production.