70 lines
3.3 KiB
Markdown
70 lines
3.3 KiB
Markdown
---
|
|
id: TASK-1
|
|
title: Add complete chat transaction context
|
|
status: Done
|
|
assignee: []
|
|
created_date: '2026-06-15 13:52'
|
|
updated_date: '2026-06-15 19:54'
|
|
labels: []
|
|
dependencies: []
|
|
priority: high
|
|
ordinal: 1000
|
|
---
|
|
|
|
## Description
|
|
|
|
<!-- SECTION:DESCRIPTION:BEGIN -->
|
|
Make the savings chat evaluate every transaction matching the selected date range, account, and basis instead of silently sampling 150/300 rows. This includes exact context totals, a full prompt context for the AI action, and regression coverage for large transaction windows.
|
|
<!-- SECTION:DESCRIPTION:END -->
|
|
|
|
## Acceptance Criteria
|
|
<!-- AC:BEGIN -->
|
|
- [x] #1 Chat context totals count all matching transactions in the selected range, including more than 400 rows
|
|
- [x] #2 Account filtering is applied before counting and summing transactions
|
|
- [x] #3 The chat action no longer accepts or passes maxTransactions and reports the complete transaction count
|
|
- [x] #4 The frontend no longer passes maxTransactions and continues to show exact totals
|
|
- [x] #5 Regression tests, lint, and build verification pass or documented blockers are reported
|
|
<!-- AC:END -->
|
|
|
|
## Implementation Plan
|
|
|
|
<!-- SECTION:PLAN:BEGIN -->
|
|
1. Add Convex test dependencies and Vitest edge-runtime config
|
|
2. Write failing convex-test regression for 450 chat transactions and account filtering
|
|
3. Refactor savingsChat backend to remove maxTransactions and use indexed full scans
|
|
4. Update schema indexes for account+basis filters
|
|
5. Remove frontend maxTransactions arguments
|
|
6. Run focused test, lint, and build verification
|
|
7. Record verification notes and leave task In Progress pending user confirmation
|
|
<!-- SECTION:PLAN:END -->
|
|
|
|
## Implementation Notes
|
|
|
|
<!-- SECTION:NOTES:BEGIN -->
|
|
RED verified: npx vitest convex/savingsChat.test.ts --run fails because getContext does not return isComplete and still uses limited context shape.
|
|
|
|
Verification:
|
|
- PASS npx vitest convex/savingsChat.test.ts --run
|
|
- PASS npm run build
|
|
- PASS npx eslint convex/savingsChat.ts convex/savingsChat.test.ts src/pages/SavingsChatPage.tsx vitest.config.ts
|
|
- BLOCKED npm run lint still fails on pre-existing/unrelated files: convex/bank/comdirectProvider.ts, convex/bank/config.ts, src/components/import/TanAwaitDialog.tsx, layout/UI fast-refresh exports, SettingsPage.tsx, and related existing warnings. SavingsChatPage is no longer in the lint output.
|
|
|
|
Review fix:
|
|
- Added effective-basis fallback for legacy transactions without effectiveMonth by reading bookingDate range and including only rows where effectiveMonth is undefined.
|
|
- Added prompt-context regression coverage via internal.savingsChat.getPromptContext.
|
|
- PASS npx vitest convex/savingsChat.test.ts --run (2 tests)
|
|
- PASS npm run build
|
|
- PASS npx eslint convex/savingsChat.ts convex/savingsChat.test.ts src/pages/SavingsChatPage.tsx vitest.config.ts
|
|
|
|
Final review:
|
|
- Subagent re-review found no remaining correctness/spec issues.
|
|
- Prior blocker resolved: effective-basis loading now includes rows without effectiveMonth via bookingDate fallback queries.
|
|
- Fresh full npm run lint still fails only on unrelated existing files; no SavingsChatPage issue remains.
|
|
<!-- SECTION:NOTES:END -->
|
|
|
|
## Final Summary
|
|
|
|
<!-- SECTION:FINAL_SUMMARY:BEGIN -->
|
|
Closed as Done after explicit user confirmation to mark all tasks done.
|
|
<!-- SECTION:FINAL_SUMMARY:END -->
|