MVP Roadmap
This page explains what must become true before the project should claim MVP completeness.
Completed
- durable replication jobs
- retry and terminal-failure handling
- guarded replication state transitions
- strict bucket/object path validation
In Progress
- streaming uploads
- upload request-size limits
Next
- database migrations and full schema constraints
- API key authentication
- observability and health endpoints
Why This Order
- Correctness first: data loss and corruption risks must be controlled before API breadth expands.
- Safety second: bounded resource use and validation reduce exploit and outage risk.
- Operability third: maintainers need evidence before they can debug incidents well.
- Surface expansion later: more APIs are useful only when the core is trustworthy.
MVP Exit Criteria
MVP is complete when:
- upload, download, and presign flows are reliable under expected load
- replication jobs are durable and observable
- auth and access control are enabled
- schema lifecycle is migration-managed
- integration tests cover both success and key failure paths
- docs let a new maintainer run and debug the service
Source: PLAN.md