---
phase: 12
plan: 02
type: execute
wave: 2
depends_on: [1]
files_modified: []
autonomous: false
requirements: [OFFER-11, OFFER-15, OFFER-16, OFFER-17, OFFER-18]
must_haves:
truths:
- "Migration 0008 (offer_macros/offer_micros additive columns + offer_tier_services table) has been applied to the production database"
- "Plan 03's query layer and Plan 04/05's UI can read/write tier_letter, public_price, offer_macros category/ticket/transformation-promise fields, and offer_tier_services against real production data"
artifacts: []
key_links:
- from: "scripts/push-12-offer-tier-schema.ts"
to: "production Postgres (178.104.27.55:54321, via SSH+docker exec)"
via: "manual operator run of the idempotent push script against prod DATABASE_URL"
pattern: "tier_letter|public_price|offer_tier_services"
---
BLOCKING checkpoint: apply migration 0008 (`src/db/migrations/0008_offer_tier_schema.sql`,
created in Plan 01) to the production database. Per CLAUDE.md Data Safety and project memory
(`project_clienthub_deploy_db.md`: Gitea→Coolify, prod Postgres reachable only via SSH+docker
exec, port 54321 firewalled from outside), this migration MUST be applied to prod BEFORE
Plans 03-05 (query layer, server actions, and UI that read/write `tier_letter`, `public_price`,
the new `offer_macros` columns, and `offer_tier_services`) are exercised against production
data.
Purpose: Unblock Wave 2/3 schema-dependent code per the project's locked migration convention
(migrations applied to prod before schema-dependent code ships).
Output: Migration 0008 applied to production; confirmation that `offer_tier_services` exists
and the new columns are queryable on the live DB.
@$HOME/.claude/get-shit-done/workflows/execute-plan.md
@$HOME/.claude/get-shit-done/templates/summary.md
@.planning/PROJECT.md
@.planning/STATE.md
@scripts/push-12-offer-tier-schema.ts
@src/db/migrations/0008_offer_tier_schema.sql
Task 1: Apply migration 0008 to production database
Plan 01 created `src/db/migrations/0008_offer_tier_schema.sql` (additive-only: 11
`ALTER TABLE ... ADD COLUMN IF NOT EXISTS` statements on `offer_macros`/`offer_micros`, a
guarded `CHECK` constraint on `offer_micros.tier_letter`, and
`CREATE TABLE IF NOT EXISTS offer_tier_services` + index) and the idempotent push script
`scripts/push-12-offer-tier-schema.ts` that applies it.
This script CANNOT be run from this environment: the production DATABASE_URL
(`178.104.27.55:54321`) is firewalled and reachable only via SSH to the production host
(per `project_clienthub_deploy_db.md`). Opening a root SSH session to the shared
production host is outside this agent's authorization.
On your machine (or via your existing SSH access to the production host), run the
migration push script against the production database, then confirm the new schema
objects exist:
1. SSH to the production host and exec into the Postgres container (same pattern used
for the Phase 11 migration apply):
```bash
ssh root@178.104.27.55
docker exec -it psql -U clienthub -d clienthub
```
OR, if you prefer running the idempotent TypeScript push script with
`DATABASE_URL` pointed at prod (via SSH port-forward / tunnel, same as Phase 11):
```bash
npx tsx scripts/push-12-offer-tier-schema.ts
```
2. If using `psql` directly, run the contents of
`src/db/migrations/0008_offer_tier_schema.sql` (copy/paste — it is additive-only and
idempotent, safe to run even if partially applied already).
3. Confirm the new columns and table exist:
```sql
\d offer_macros
\d offer_micros
\d offer_tier_services
```
Expect: `offer_macros` has `description`, `category`, `ticket`, `is_archived`,
`cliente_ideale`, `risultato`, `tempo`, `pain`, `metodo`; `offer_micros` has
`tier_letter` (with a CHECK constraint `offer_micros_tier_letter_check`) and
`public_price`; `offer_tier_services` table exists with columns `tier_id`,
`service_id`, `created_at` and a composite primary key on (`tier_id`, `service_id`).
4. Confirm no existing data was affected:
```sql
SELECT count(*) FROM offer_macros;
SELECT count(*) FROM offer_micros;
SELECT count(*) FROM clients;
SELECT count(*) FROM projects;
```
Row counts should match pre-migration counts (migration is additive — no rows
deleted).
Type "applied" once migration 0008 has been run against production and you've confirmed the new columns/table exist (or describe any errors encountered).
## Trust Boundaries
| Boundary | Description |
|----------|-------------|
| Operator (human) -> Production Postgres | Manual SSH+docker exec session per project convention; only additive, idempotent SQL from migration 0008 is executed |
## STRIDE Threat Register
| Threat ID | Category | Component | Disposition | Mitigation Plan |
|-----------|----------|-----------|-------------|-----------------|
| T-12-05 | Denial of Service / Tampering | Manual prod migration apply | accept | Operator runs only the additive, idempotent migration 0008 (already grep-gated in Plan 01 against DROP/TRUNCATE/RENAME); per project convention this is a manual, out-of-band step not reachable via any HTTP endpoint |
1. Operator confirms (via `\d offer_macros`, `\d offer_micros`, `\d offer_tier_services` on prod) that all Plan 01 additive columns and the new junction table exist.
2. Operator confirms row counts for `offer_macros`, `offer_micros`, `clients`, `projects` are unchanged (additive migration, no data loss).
3. Resume-signal "applied" received before Wave 3 (Plans 04/05) begins relying on these columns/table against production data.
- Migration 0008 is live on production.
- `offer_tier_services` table and all new `offer_macros`/`offer_micros` columns are queryable on prod.
- No row in `clients`, `projects`, `payments`, `phases`, `offer_macros`, or `offer_micros` was deleted or modified destructively.