docs(15-01): complete phishing triage schema foundation plan

- Migration 097 applied and verified in dev DB (7 tables, idempotent re-run confirmed)
- SUMMARY.md documents schema design decisions and worktree-path caveat with scripts/apply-migrations.sh
This commit is contained in:
lorentz 2026-07-15 07:36:29 -04:00
parent 84a37e20be
commit 15d8a691a7

View file

@ -0,0 +1,92 @@
---
phase: 15-data-model-detection-ticket-evidence
plan: 01
subsystem: database
tags: [postgres, migration, schema, phishing-triage]
# Dependency graph
requires: []
provides:
- "migrations/097_phishing_triage_schema.sql applied to dev DB"
- "7-table phishing-triage schema: campaigns, reports, messages, indicators, classifications, remediation_actions, audit_events"
- "reports table fully designed with ticket_id FK, content_hash idempotency key, matched_patterns/evidence JSONB columns, uq_reports_ticket_id unique constraint"
affects: [15-02-phishing-detector, 16-message-parsing, 18-campaign-grouping, 19-classification, 20-remediation]
# Tech tracking
tech-stack:
added: []
patterns:
- "Schema-only foundation migration lands before any service writes to it (mirrors migration 091 pax8 pattern)"
- "UUID PKs via gen_random_uuid() (no extension needed, Postgres 16 built-in)"
- "Stub tables for future-phase entities created now with hard FKs where insert order is guaranteed within the same migration"
key-files:
created: [migrations/097_phishing_triage_schema.sql]
modified: []
key-decisions:
- "reports.ticket_id is a hard FK to tickets(id) with UNIQUE constraint (uq_reports_ticket_id) so the Plan 02 detector can upsert one report per ticket via ON CONFLICT (ticket_id)"
- "campaigns/messages/indicators/classifications/remediation_actions/audit_events created as stubs now (per STATE.md decision: durable schema lands in Phase 15, before any service that writes to it) so Phases 16-21 never need a second foundation migration"
- "Applied migration directly via docker exec psql (not scripts/apply-migrations.sh) because that script hardcodes the main-repo path /opt/stacks/pulse/migrations, which does not yet contain this file while running inside a worktree"
patterns-established:
- "Pattern: schema-only migrations for future phases mirror migrations/091_pax8_tables.sql header framing (numbered list of tables, explicit note on what's populated now vs. stubbed)"
requirements-completed: [DETECT-01, DETECT-02, EVID-01]
# Metrics
duration: 12min
completed: 2026-07-15
---
# Phase 15 Plan 01: Phishing Triage Schema Foundation Summary
**Migration 097 lays down the full 7-table phishing-triage schema in Postgres — reports table fully designed for ticket_id, content_hash idempotency, matched_patterns, and EVID-01 evidence; six other tables stubbed for Phases 16-21 — applied and verified in the dev DB.**
## Performance
- **Duration:** 12 min
- **Started:** 2026-07-15T11:23:00Z
- **Completed:** 2026-07-15T11:35:46Z
- **Tasks:** 2 completed
- **Files modified:** 1
## Accomplishments
- Created `migrations/097_phishing_triage_schema.sql` with all 7 phishing-triage tables (campaigns, reports, messages, indicators, classifications, remediation_actions, audit_events), each `CREATE TABLE IF NOT EXISTS`, snake_case columns, UUID PKs via `gen_random_uuid()`.
- `reports` table fully designed: `ticket_id BIGINT NOT NULL REFERENCES tickets(id)`, `content_hash TEXT NOT NULL` (D-04 idempotency), `matched_patterns JSONB NOT NULL DEFAULT '[]'`, `evidence JSONB NOT NULL DEFAULT '{}'`, `campaign_id UUID REFERENCES campaigns(id)` (nullable), and `CONSTRAINT uq_reports_ticket_id UNIQUE (ticket_id)`.
- Applied migration to the running dev Postgres (`pulse-postgres` container) via `docker exec -i pulse-postgres psql`.
- Verified all 7 tables exist in `information_schema.tables`, confirmed `reports` columns/constraint via `\d reports`, and re-ran the migration to confirm full idempotency (all statements returned `NOTICE: ... already exists, skipping`, zero errors).
## Task Commits
1. **Task 1: Write migration 097 — 7-table phishing triage schema** - `84a37e2` (feat)
2. **Task 2: Apply migration 097 to the dev database and verify tables exist** - no commit (DB-only verification step; no file changes produced — the migration file was already committed in Task 1)
**Plan metadata:** (this SUMMARY.md commit)
## Files Created/Modified
- `migrations/097_phishing_triage_schema.sql` - 7-table phishing-triage schema: campaigns, reports (fully designed for Plan 02's detector), messages, indicators, classifications, remediation_actions, audit_events (all stubs except reports)
## Decisions Made
- Hard FK + UNIQUE constraint on `reports.ticket_id` (rather than a soft ref like PAX8's company/product tables) because report inserts always happen one-at-a-time against an already-synced ticket — no batch-insert-order ambiguity to avoid, unlike PAX8's multi-entity sync passes.
- Applied the migration directly via `docker exec -i pulse-postgres psql -U pulse_user -d pulse_autotask < migrations/097_phishing_triage_schema.sql` instead of `scripts/apply-migrations.sh`, because that script hardcodes `MIGRATIONS_DIR="/opt/stacks/pulse/migrations"` (the main repo checkout) — inside this worktree the new file doesn't exist at that path yet, so the script would silently fail to find it. Documented here for the orchestrator/user; no change made to the script itself (out of scope for this plan).
## Deviations from Plan
None - plan executed exactly as written. Task 2's verification uncovered a worktree-path caveat with `scripts/apply-migrations.sh` (documented above under Decisions Made) but this did not require any code change — the plan's own fallback instruction ("Otherwise pipe the file into the container's psql") was used as designed.
## Issues Encountered
None.
## User Setup Required
None - no external service configuration required. Migration applied directly to the existing dev Postgres container; no new env vars or credentials needed.
## Next Phase Readiness
- All 7 phishing-triage tables exist in the dev database and are idempotently re-appliable.
- `reports` table is fully ready for Plan 02's phishing-detector service to upsert into via `ON CONFLICT (ticket_id)`.
- `campaigns`, `messages`, `indicators`, `classifications`, `remediation_actions`, `audit_events` are schema-ready stubs for Phases 16-21 — no second foundation migration will be needed.
- No blockers.
---
*Phase: 15-data-model-detection-ticket-evidence*
*Completed: 2026-07-15*