Roadmap

Phase T0: License And Dependency Inventory

• Keep the repository Apache-2.0.
• Maintain a dependency allowlist.
• Block AGPL Studio and CLI components.
• Run tests and secret scans before any remote push.
• Publish the bootstrap repository to GitHub before adding worker dependencies.

Phase T1: Product And Consent Design

• Define training consent.
• Define dataset preview contracts.
• Define redaction and exclusion rules.
• Define artifact retention and deletion policy.
• Define the server-side Scooling-to-Scooling-Lab authorization envelope.

Phase T2: Training Job API Contract

• Add job create, status, cancel, and artifact metadata contracts.
• Keep validation server-side.
• Preserve audit metadata without learner-content logs.
• Add content-free fixtures that can be shared with Scooling and Knowtation tests.
• Add content-free provenance records, bounded retention policies, idempotent deletion, and deletion verification before private data or real artifacts are allowed.

Phase T3: Worker Skeleton

• Add an isolated worker process.
• Run only approved dependencies.
• Store artifacts outside the main app database.
• Keep real learner data and private model artifacts out of repository history.
• Keep the fixture worker path dependency-free until the isolation step is accepted.

Slice 5: Fixture Provenance, Retention, And Deletion

• Emit validated provenance for completed fixture jobs.
• Evaluate retention on access and through explicit sweeps.
• Delete artifact placeholders, metadata, provenance, and job content fields as one cascade.
• Verify deleted artifact hashes are absent from store outputs before private data is allowed.