Replace @vercel/postgres, Blob, and Upstash with lib/sql.js, MinIO object
storage, and CT 102 Redis rate limits. Add Dockerfile for Dokploy deploy,
homelab runbooks, Neon data-copy helper, and point CI smoke/visual at the
homelab URL instead of Vercel previews.
Co-authored-by: Cursor <cursoragent@cursor.com>
Convoy: single-sql-client (P1 quality, launch sequence step 8)
Addresses: AGENTS.md Gotcha #1, .convoys/ship-readiness.md P1 #8
## Decisions
- D1: Caller inventory = 2 files (1 source + 1 test), not "~3 based on graph".
Only pages/api/auth-utils.js imports `db`; test/api/auth-utils.test.js mocks
it purely to satisfy the import graph (the 5 tests exercise
generateToken/verifyToken, not isAdmin/getUserById).
- D2: Migrate both call sites (isAdmin, getUserById) to @vercel/postgres
tagged-template SQL. Queries are SELECT-only, single-table,
single-numeric-parameter — byte-equivalent translation; same result shape
({rows, rowCount}); no transaction or pool semantics differ.
- D3: KEEP @neondatabase/serverless as a dep. 11 scripts/* files still use
`neon()` directly (setup-neon-db.js, migrations/, reset-db.js, 8 historical
add-*/fix-*/seed-* jobs). They are out of scope per the no-go-zones rule
and the convoy spec; purging the dep entirely would be its own convoy
(queued as `purge-neondatabase-serverless-fully`, blocked on migration-tool).
- D4: sql.unsafe audit — NOT a real injection vector with current callers
(userId comes from a verified JWT, is a numeric SERIAL id). Security
finding: NO. Pure refactor + foot-gun removal that prevents the FUTURE
caller that would have been the incident.
- D5: Test mock cleanup — drop the now-unneeded `vi.mock('../../lib/database.js')`
call + unused `vi` import. Test count + assertions unchanged (5/5).
## Per-file changes
- pages/api/auth-utils.js: swap `import { db } from '../../lib/database.js'`
for `import { sql } from '@vercel/postgres'`; rewrite isAdmin's
`db.query(SELECT … WHERE id = $1, [userId])` and getUserById's same shape
to `sql\`SELECT … WHERE id = ${userId}\``. Same try/catch, same
result.rows[0] access, same error returns.
- test/api/auth-utils.test.js: drop vi.mock for lib/database.js + the unused
`vi` import. 5/5 tests still pass.
- lib/database.js: DELETED (47 lines removed; manual-interpolation + sql.unsafe
wrapper is gone).
- .convoys/single-sql-client.md: NEW (the convoy file documenting all
decisions + caller inventory + verification + risks + follow-ups).
## Verification
- npm run lint → 128 problems (baseline preserved, no regression)
- npm run test:run → 21/21 pass (vitest)
- Grep "lib/database" --type js -l → 0 hits anywhere
- Grep "@neondatabase/serverless" --type js -l → still matches the 11
scripts/* sites (expected; out of scope per D3)
- node --check pages/api/auth-utils.js → exit 0
## Scope note
This convoy collapses the lib/database.js abstraction onto the canonical
@vercel/postgres surface for pages/api/**. It does NOT eliminate
@neondatabase/serverless from the dependency tree — that would require
migrating the scripts/* helpers, which is out of scope here (no-go-zones
rule + convoy spec). Queued as a follow-up.
## Live smoke
Deferred. The two migrated functions (isAdmin, getUserById) are only
reachable via pages/api/admin/index.js which requires an admin Bearer
token and a populated users table in prod Neon. Byte-equivalent SQL +
identical result shape gives high confidence; rollback is a single-commit
revert if a post-merge admin action 500s.
Co-authored-by: Cursor <cursoragent@cursor.com>
Closes AGENTS.md gotcha #11 (well, the relevant half of it — "Testing:
None yet" line in §6 is now stale).
Installs vitest@^3.2.4 (single devDep, no UI / coverage / jsdom) and
adds 16 unit tests across 3 files that lock in post-Brief-1/2/4
behavior:
test/lib/auth-secret.test.js (3 tests)
- JWT_SECRET exports the env value
- JWT_TOKEN_TTL is canonical 24h
- Module throws at load when JWT_SECRET is empty
test/lib/permission-middleware.test.js (8 tests)
- getUserFromRequest returns null for: missing header, non-Bearer
scheme, malformed token, wrong-secret token, expired token,
valid-token-no-user-row
- Returns user object for valid token + user row
- Brief 2 regression lock: does NOT return the synthetic admin
shape { userId: 1, email: 'admin@tcgvault.com', role: 'admin' }
when no Authorization header is present
test/api/auth-utils.test.js (5 tests)
- generateToken issues 24h JWT (exp - iat === 86400)
- Payload includes userId, email, role
- verifyToken round-trips valid tokens
- Returns null for malformed / wrong-secret tokens
CI: re-enabled the previously commented-out test: job in
.github/workflows/ci.yml. Blocking (no || true wrapper) — vitest is
the first runner in this repo and we want CI red on test regression.
JWT_SECRET is set via a CI-only fake; production secret is unaffected.
Rate-limit (Brief 4) coverage deferred to a future expand-auth-tests
convoy per architect's call (R11). package.json has "type": "module"
so vitest's default Vite-based transform handles .js ESM out of the
box — no transform config needed.
Convoy: fix-auth-bypass / Brief 5 (last brief)
Co-authored-by: Cursor <cursoragent@cursor.com>
