chore: add claude command to resolve drizzle merge conflicts

.claude/commands/resolve-drizzle-migrations.md

@@ -0,0 +1,56 @@
+---
+allowed-tools: Bash(git show:*), Bash(git checkout *:*), Bash(git status:*), Bash(git diff:*), Bash(git log:*), Bash(cd server && pnpm drizzle-kit generate:*), Bash(rm server/src/migration:*), Bash(ls:*), Bash(python3:*)
+description: Resolve Drizzle migration conflicts by accepting upstream and regenerating local migrations
+argument-hint: Upstream branch (default: main)
+---
+
+## Context
+
+- Current branch: !`git branch --show-current`
+- Merge state: !`git status --short server/src/migration/db/sql/`
+
+## Your task
+
+Resolve Drizzle ORM migration conflicts between the current branch and an upstream branch.
+The upstream branch's migrations keep their original indices. The current branch's schema
+changes are collapsed into a single freshly-generated migration at the next available index.
+
+**Upstream branch:** `$ARGUMENTS` (default to `main` if empty).
+
+### Steps
+
+1. **Get both journals.** Read the upstream journal and the current branch's journal:
+   - Upstream: `git show <upstream>:server/src/migration/db/sql/meta/_journal.json`
+   - Ours: if mid-merge and the file is conflicted, use `git show HEAD:server/src/migration/db/sql/meta/_journal.json`. Otherwise read it from the working tree.
+
+2. **Find the divergence point.** Walk both entry lists and find the last entry where `tag` matches at the same `idx`. Everything after that is either upstream-only or ours-only.
+
+3. **Identify our branch-only migrations.** These are entries in our journal whose `tag` does not appear in the upstream journal. Record each one's:
+   - SQL file: `server/src/migration/db/sql/<tag>.sql`
+   - Snapshot: `server/src/migration/db/sql/meta/<idx>_snapshot.json`
+
+4. **Accept upstream's migration state.** For each upstream-only entry (past the shared base), plus the journal itself:
+   ```
+   git checkout <upstream> -- server/src/migration/db/sql/<tag>.sql
+   git checkout <upstream> -- server/src/migration/db/sql/meta/<idx>_snapshot.json
+   git checkout <upstream> -- server/src/migration/db/sql/meta/_journal.json
+   ```
+
+5. **Remove our branch-only artifacts.** Delete the SQL and snapshot files identified in step 3:
+   ```
+   rm server/src/migration/db/sql/<tag>.sql
+   rm server/src/migration/db/sql/meta/<idx>_snapshot.json
+   ```
+
+6. **Regenerate.** Run `cd server && pnpm drizzle-kit generate`. This diffs the current schema TypeScript against upstream's latest snapshot and produces a single correct migration.
+
+7. **Verify.**
+   - Read the updated `_journal.json` — confirm the new entry follows upstream's last entry.
+   - Spot-check that the new snapshot's `prevId` matches the last upstream snapshot's `id`.
+   - List the SQL directory and confirm no duplicate-numbered files or orphans.
+
+8. **Report.** Summarize:
+   - Migrations accepted from upstream (tags)
+   - Migrations removed from our branch (tags)
+   - New migration generated (index, tag, file path)
+   - Remind the user to review the generated SQL before committing.