A per-slice historical record of how blockdata is built — one page per implementation step/slice. This complements generated Contracts: contracts are the governed rules, while the Implementation Log is the durable, append-only journal of each slice as it ships.
Page structure
Every slice page follows the same 10 sections:
- Summary — what the slice delivered (1–2 lines)
- Scope — in / explicitly out
- Source / reference — what it was based on (legacy reference, plan doc)
- Pages & routes — new / changed / bypassed-or-removed (+ redirects)
- Dependencies — npm + Python added, allowlist entries + why
- Data & migrations — tables + Alembic migration(s) + identity
- API — endpoints added / changed (verb · path · auth)
- Contracts — governance JSON(s) touched + rules locked
- Verification — gates, smoke, live URL / evidence
- Residuals & links — deferred items + plan doc / commits
Slices
| # | Slice | Status |
|---|---|---|
| 01 | Auth (login / register) | page pending |
| 02 | Onboarding wizard (organization → members → project) | planned |
Stack comparison: blockdata vs. AI-agent template
Layer-by-layer — what blockdata adopts, keeps, swaps, or still has to decide if the vstorm full-stack-ai-agent-template's Python backend becomes the base.
Graph DB — when it earns its keep
Scenario-grounded analysis of when a graph database (AGE / Neo4j / ArangoDB) pays off vs when plain Postgres + vector is enough, for the book editor + BlockData + agent products.