Merge pull request 'docs(ai): refresh handoff state after PR #153 merge' (#154) from chore/post-153-handoff into main

Reviewed-on: #154

.ai/CURRENT_TASK.md

@@ -1,22 +1,20 @@
 # CURRENT_TASK.md
 
-**Task:** Restore a fully green CI gate on `main` and lock it via branch protection so future merges can't introduce silent rot.
+**Task:** No active task — pick from [`TODO.md`](TODO.md).
 
-**Status:** in-progress
+**Status:** ready for next pickup.
 
-**Definition of Done:**
-- [ ] PR #150 (`fix/ci-workflow-config`) merged. Both `CI / backend (pull_request)` and `CI / frontend (pull_request)` show success on the merge commit.
-- [ ] `CI / backend (pull_request)` added to required status checks on `main` in Gitea branch protection (frontend is already required).
-- [ ] The 54 real backend test failures (left after #149's infra cleanup) categorized and fixed in a follow-up PR. Target: 0 failures, 0 errors on a `pytest` run inside `resolutionflow_backend`.
-- [ ] `npm run lint` stays at 0 errors after the cleanup PR (already at 0 on main).
-- [ ] Append a SESSION_LOG.md entry summarizing what shipped.
+## Recommended next moves
 
-**Assumptions:**
-- The 54 failures fall into a small number of root-cause categories (likely 3–5: fixture-scoping leaks, DB cleanup ordering, account_id propagation in test seed paths). Verify before assuming.
-- The pytest-asyncio 0.24 + pytest 8.4 toolchain bumped in #149 is the right baseline; do not revert.
-- `DATABASE_TEST_URL` is the only DB URL conftest will honor; do not weaken the safety guard added in `dab740d`.
+1. **Promote `CI / e2e (pull_request)` to required on `main`.** Two consecutive PR runs (#150 and #153) have now finished green on the e2e job. That was the threshold the prior CI-recovery task set for promoting it. Branch protection update only — no code change.
+2. **Pick a backlog item.** Top of `TODO.md` "Up next" is the `data-testid` e2e-stability work (PR #152 spent five one-line selector updates chasing UI churn — adding stable test IDs to a small set of high-value elements would make those tests immune to copy/route renames). The new `currentChatRef` silent-return audit added in #153's session is in Backlog and is a natural pairing with the bug fix that was just shipped.
 
-**Out of scope:**
-- New feature work on FlowPilot (Phase 10+) or PSA — keep this branch focused on CI debt.
-- Frontend lint warnings (23 remain after #149; they're missing-deps in useEffect, opt-in cleanup later).
-- RLS test suite (`test_rls_isolation.py`) — gated behind `RUN_RLS_TESTS=1` and not in the default CI run.
+## Previous task — closed out
+
+**Task:** Land PR #153 — fix the `AssistantChatPage` prefill `currentChatRef` bug that silently dropped AI follow-up responses in the task lane.
+
+**Status:** complete (2026-04-26).
+
+- PR #153 merged as commit `68fcdc6` on `main`. Backend, frontend, and e2e all green on the merged SHA after the env-var fix.
+- E2e CI needed a stub `ANTHROPIC_API_KEY` in the workflow so the AI-gated `POST /api/v1/ai-sessions` endpoint stops returning 503; the Playwright `page.route` stub still intercepts the actual `/chat` call in the browser, so no real Anthropic traffic occurs.
+- Regression test `frontend/e2e/assistant-chat-prefill.spec.ts` is part of the e2e suite going forward.

.ai/HANDOFF.md

@@ -2,62 +2,27 @@
 
 # HANDOFF.md
 
-**Last updated:** 2026-04-25 06:12 EDT
+**Last updated:** 2026-04-26 04:55 EDT
 
-**Active task:** Restore green CI gate on `main` and lock it via branch protection. See [CURRENT_TASK.md](CURRENT_TASK.md).
+**Active task:** None — pick from [`TODO.md`](TODO.md). See [`CURRENT_TASK.md`](CURRENT_TASK.md) for recommended next moves.
 
-**Branch:** `fix/ci-workflow-config`
+**Branch:** `main` is the home position. Recent merges: PR #150 (CI recovery, `87bb20b`), PR #153 (prefill `currentChatRef` fix, `68fcdc6`).
 
-## Current state
+## Where things stand
 
-Previous session fixed the 54 real backend failures left after #149. The default backend suite is now green locally:
-
-```bash
-docker exec resolutionflow_backend bash -lc 'pytest --override-ini="addopts=" -q > /tmp/full-backend.log 2>&1; code=$?; tail -n 160 /tmp/full-backend.log; exit $code'
-# 1076 passed, 35 deselected in 1347.41s (0:22:27)
-```
-
-Targeted validation also passed:
-
-- `tests/test_session_resolutions_api.py tests/test_session_sharing.py tests/test_session_suggested_fixes_api.py tests/test_survey.py tests/test_tenant_isolation_p0.py tests/test_tree_sharing.py tests/test_trees.py::TestTrees::test_delete_tree_cleans_up_folder_and_tag_assignments tests/test_uploads.py::test_delete_upload_forbidden_for_non_owner` → `73 passed`
-- PDF export tests → `3 passed`
-- Prompt/PSA/resolution/script-builder subset → `14 passed`
-- Admin/AI/branch subsets → `11 passed`
-
-## What changed
-
-Production fixes:
-
-- CI/backend dev image now installs WeasyPrint system libraries.
-- Public share-token and survey routes are mounted outside tenant auth; protected share management remains tenant-protected.
-- Folder creation now persists `UserFolder.account_id`.
-- Script Builder save-to-library now persists `ScriptTemplate.account_id`.
-- Resolution output generation eager-loads `AISession.steps` to avoid async lazy-load `MissingGreenlet`.
-- AI session model now declares the generated `search_vector` column already present in Alembic, so `create_all` test schemas match runtime migrations.
-- Direct account-role update now rejects `"owner"`; ownership changes must use the transfer path.
-- Assistant prompt marker examples no longer include a literal executable `create_spin_off_ticket` payload.
-
-Test/harness fixes:
-
-- Test seeds updated for tenant-scoped `account_id` columns on sessions, branches, resolution outputs, script templates, PSA connections, folders, schedules, and categories.
-- Tests aligned with 404-not-403 resource-hiding policy.
-- Disabled-AI tests now restore both Anthropic and Google key settings.
-- Pytest harness closes pytest-asyncio's leftover clean loop and ignores known unclosed asyncio/asyncpg teardown ResourceWarnings that otherwise appear at arbitrary later setup points under `filterwarnings = error`.
+- CI is healthy on `main`: backend, frontend, and e2e all green on the latest commits.
+- Branch protection on `main`: PR-only merges, force-push blocked, **`CI / frontend (pull_request)` required**, **`CI / backend (pull_request)` required**, `CI / e2e (pull_request)` not yet required.
+- Two consecutive PR runs (#150, #153) finished green on e2e. The "promote e2e to required" gate from the prior task is now satisfiable.
+- Backend AI-gated endpoints (`POST /ai-sessions`, `/chat`, `/respond`, etc.) call `_require_ai_enabled()` and return 503 if no provider key is set. The e2e CI job now sets a stub `ANTHROPIC_API_KEY` so any future test that exercises those flows can rely on it; tests should still stub the actual AI calls in the browser via `page.route` so no real Anthropic traffic occurs.
 
 ## Immediate next steps
 
-1. Commit current working tree if not already committed with trailer:
-   `Co-Authored-By: Codex <noreply@openai.com>`.
-2. Check PR #150 status on Gitea. If both `CI / backend (pull_request)` and `CI / frontend (pull_request)` are green, merge it.
-3. After #150 merges, add `CI / backend (pull_request)` to required status checks on main:
-   ```bash
-   PATCH /repos/chihlasm/resolutionflow/branch_protections/main
-   { "status_check_contexts": ["CI / frontend (pull_request)", "CI / backend (pull_request)"] }
-   ```
-   `$GITEA_TOKEN` is in `.claude/settings.local.json`.
-4. Run/confirm frontend lint if needed for the final DoD item (`npm run lint` was already green after #149, but this session did not rerun it).
+1. (Optional, ops-only) Promote `CI / e2e (pull_request)` to required on `main` in Gitea branch protection.
+2. Pick the next backlog item from `TODO.md`. Top of "Up next" is the `data-testid` e2e-stability audit; the new `currentChatRef` silent-return audit (added to backlog in this session) is a natural pairing with the bug fix that just shipped.
 
-## Open questions
+## Useful breadcrumbs
 
-- PR #150 was not rechecked or merged in this session.
-- Branch protection was not updated in this session.
+- The fix that just landed: [`frontend/src/pages/AssistantChatPage.tsx`](../frontend/src/pages/AssistantChatPage.tsx) — `currentChatRef.current = session.session_id` after `setActiveChatId` in the dashboard prefill effect.
+- Regression test: [`frontend/e2e/assistant-chat-prefill.spec.ts`](../frontend/e2e/assistant-chat-prefill.spec.ts).
+- E2e env convention: [`.gitea/workflows/ci.yml`](../.gitea/workflows/ci.yml) — `ANTHROPIC_API_KEY` is stubbed in the e2e job env. Tests that exercise AI-gated endpoints should stub the actual AI calls in the browser, not rely on a real key.
+- Silent-return follow-up entry: [`.ai/TODO.md`](TODO.md), Backlog section.

.ai/SESSION_LOG.md

@@ -12,6 +12,40 @@
 
 ---
 
+## 2026-04-26 03:50 EDT — Claude Code — Ship AssistantChatPage prefill `currentChatRef` fix; close out PR #150
+
+- User reported a troubleshooting-session bug: after answering a subset of task-lane questions and clicking *Send N of M Responses*, no AI response appeared. Traced to `AssistantChatPage`: the dashboard prefill effect set `activeChatId` after creating a new chat session but never updated `currentChatRef.current`. The `currentChatRef.current !== sentForChatId` guard in `handleSend` and `handleTaskSubmit` then bailed silently on every later request and discarded the AI's reply. The user message was already pushed to the chat before the await, so the user saw their answers but nothing else.
+- Fix: one-line addition mirroring `handleNewChat` and `handleResumeNew` — assign `currentChatRef.current = session.session_id` immediately after `setActiveChatId(session.session_id)` in the prefill effect. Branched off `origin/main` as `fix/tasklane-prefill-ref`; PR #153 opened on Gitea.
+- Authored a Playwright regression test `frontend/e2e/assistant-chat-prefill.spec.ts` that drives the real dashboard prefill flow against the real backend, stubs `/ai-sessions/*/chat` with `page.route` for deterministic turn-1/turn-2 responses, and asserts the second AI message renders. Confirmed the test fails on unfixed code at the exact assertion (`Got it — based on your answer…` never appears) and passes once the fix is restored.
+- Verified locally inside `mcr.microsoft.com/playwright:v1.58.2-noble` against the running dev stack: new spec passes, adjacent `flowpilot-chat` spec still passes, `tsc -b` clean. `resume.spec` and `history.spec` failures observed are pre-existing real-backend fixture collisions, unrelated to this change.
+- First CI run on PR #153 failed on infrastructure issues already addressed by PR #150: backend hit `Bind for 0.0.0.0:5432 failed: port is already allocated`, frontend hit `actions/upload-artifact@v4 not supported on GHES`. PR #150 was already merged (commit `87bb20b` on `main`). Rebased `fix/tasklane-prefill-ref` onto new `main` (force-push `1a8cb06` → `1559feb`), resolved a `.ai/TODO.md` conflict by keeping both backlog item sets, kicked off CI on the rebased SHA.
+- Confirmed `CI / backend (pull_request)` is now in branch protection's required-status-checks list (added during PR #150 close-out). `CI / e2e (pull_request)` left as not-required pending one more clean PR run as the threshold.
+- Recorded the broader silent-return concern in TODO backlog: the `currentChatRef.current !== sentForChatId` guard is applied across `handleSend`, `handleTaskSubmit`, `selectChat`, `refreshFacts`, `refreshActiveFix`, and `refreshPreview`. PR #153 fixes one symptom but the same pattern can mask other drift. Either log a Sentry breadcrumb on the mismatch path or distinguish "expected stale" (chat switch) from "unexpected stale" (ref never updated) so the latter alerts.
+- First CI run on the rebased SHA passed backend and frontend but failed e2e: the new prefill regression test couldn't render the task-lane question text. Diagnosed via the job log: `POST /api/v1/ai-sessions` calls `_require_ai_enabled()` and returns 503 when no provider key is set. The e2e CI job had neither `ANTHROPIC_API_KEY` nor `GOOGLE_AI_API_KEY` in env. Locally the dev backend has a real key, hence the local pass. The Playwright `page.route` stub on `/chat` was correct but never had a chance to fire because the upstream session-creation call was 503-ing.
+- Fix: added a stub `ANTHROPIC_API_KEY: ci-stub-key-not-used-by-tests` to the e2e job env in `.gitea/workflows/ci.yml`. The Playwright stub still intercepts the actual `/chat` call in the browser, so the backend never contacts Anthropic — the gate just needs to clear. Documented the convention in a workflow comment so future AI-touching e2e tests know what to expect. Pushed `11fe32f`; CI went all-green.
+- Merged PR #153 as `68fcdc6` on `main`. Local feature branch and remote both deleted via Gitea's `delete_branch_after_merge`.
+- Opened a small follow-up `chore/post-153-handoff` PR to refresh the now-stale `.ai/` files (this entry, plus `CURRENT_TASK.md` rolling forward to "no active task — pick from `TODO.md`" and `HANDOFF.md` updating to the post-merge home position). The `data-testid` audit at the top of `TODO.md` "Up next" or the `currentChatRef` silent-return audit added in this session's backlog are the natural next pickups.
+- Files touched: `frontend/src/pages/AssistantChatPage.tsx` (the one-line fix + comment), `frontend/e2e/assistant-chat-prefill.spec.ts` (new regression test), `.gitea/workflows/ci.yml` (stub `ANTHROPIC_API_KEY` for e2e), `.ai/TODO.md` (silent-return follow-up entry, plus conflict resolution preserving PR #150's backlog additions), `.ai/CURRENT_TASK.md`, `.ai/HANDOFF.md`, `.ai/SESSION_LOG.md` (this entry).
+
+## 2026-04-25 16:41 EDT — Codex — Stabilize PR #150 e2e selectors
+
+- Investigated the remaining PR #150 failure after backend and frontend CI were green. The e2e resume smoke test was not failing because of product behavior; it used `.bg-card` plus text filtering and matched the tree filter `<select>` before the intended session card.
+- Added stable test IDs to flow session, tree, and share cards, then updated affected e2e tests to target those cards instead of Tailwind class names.
+- Hardened the CI workflow by making Postgres healthchecks authenticate as `postgres` and baking `VITE_API_URL="${PLAYWRIGHT_API_ORIGIN}"` into the e2e frontend build.
+- Verified with `git diff --check`, frontend build in Docker, no remaining `.bg-card` e2e selectors, and focused Playwright runs in an Actions-like Ubuntu container: resume spec passed, then history/library/library-start/resume/shares passed (`6 passed`).
+- Left for next session: push this WIP commit to PR #150, watch CI, merge when all three jobs are green, then enable backend branch protection and consider the e2e gate after a reliable green run.
+- Files touched: `.gitea/workflows/ci.yml`, `frontend/e2e/history.spec.ts`, `frontend/e2e/library-start.spec.ts`, `frontend/e2e/library.spec.ts`, `frontend/e2e/resume.spec.ts`, `frontend/e2e/shares.spec.ts`, `frontend/src/components/library/TreeGridView.tsx`, `frontend/src/components/library/TreeListView.tsx`, `frontend/src/pages/MySharesPage.tsx`, `frontend/src/pages/SessionHistoryPage.tsx`, `.ai/HANDOFF.md`, `.ai/CURRENT_TASK.md`, `.ai/SESSION_LOG.md`.
+
+## 2026-04-25 12:00 America/New_York — Claude Code — Mock final AI-provider test, cache CI deps, parallelize backend with pytest-xdist
+
+- Diagnosed why CI was still red despite Codex's local 1076 passed: a single test (`test_record_decision_persists_and_bumps_state_version`) needed `ANTHROPIC_API_KEY` because the `decision: draft_template` path calls `TemplateExtractionService` → AI provider. Patched `_extract_template_parameters` with an `AsyncMock` so the test no longer depends on AI availability. Verified.
+- Pushed Codex's WIP commit `49f8856` to PR #150 (had been local-only per handoff protocol).
+- PR #150 (`fix/ci-workflow-config`) extended with cheap CI wins: `actions/cache@v3` for pip + npm in all three jobs; dropped `--cov-report=term-missing` (the custom display step parses JSON); added `--maxfail=10` so structural breakage exits fast.
+- PR #151 (`fix/ci-pytest-xdist`) opened, stacked on #150: pytest-xdist with per-worker DB isolation. `conftest.py` reads `PYTEST_XDIST_WORKER`, computes a per-worker DB URL like `…_gw0`, and synchronously CREATEs the DB on first import. The per-test `DROP SCHEMA public CASCADE` then operates on the worker's isolated DB. Verified locally: backend suite went from 22m 27s serial → 4m 28s parallel (8 workers), 1076 passed in both cases. ~5× speedup.
+- Decided NOT to do per-test transactional rollback (bigger refactor); captured for future TODO consideration.
+- Left for next session: watch CI on both PRs, merge in order (#150 first, #151 second), then enable `CI / backend (pull_request)` as a required status check on main.
+- Files touched: `backend/tests/test_session_suggested_fixes_api.py`, `backend/tests/conftest.py`, `backend/requirements-dev.txt`, `.gitea/workflows/ci.yml`, `.ai/HANDOFF.md`, `.ai/CURRENT_TASK.md`, `.ai/TODO.md`.
+
 ## 2026-04-25 06:12 EDT — Codex — Fix backend suite to green
 
 - Fixed the real backend failures left after the CI-infra cleanup: tenant-scoped seed drift, missing production `account_id` writes, public route mounting for survey/share links, Script Builder library saves, resolution output async loading, AI search schema metadata, disabled-AI fixture leakage, and prompt marker guardrails.

.ai/TODO.md

@@ -5,9 +5,13 @@
 
 ## Up next
 
-- [ ] **Parallelize backend pytest with pytest-xdist.** Currently the backend suite takes ~22 min wall-clock for `1076 passed, 35 deselected` (verified locally 2026-04-25). With `-n auto` on the homelab Gitea Actions runner, this should land in the 3–6 min range depending on core count. Blocker: `test_db` fixture in `backend/tests/conftest.py` does `DROP SCHEMA public CASCADE` per test, which two workers would race on. Standard fix: one database per worker, derived from `PYTEST_XDIST_WORKER` env var inside conftest. The runner has spare CPU, so prioritize once main is green and the 54-failure cleanup has landed.
+- [ ] **Parallelize backend pytest with pytest-xdist.** ✅ landing as PR #151. Verified locally: backend suite 22 min → 4m 28s with `-n auto` on the 8-core homelab runner. Per-worker DB isolation via `PYTEST_XDIST_WORKER` in conftest.py.
 
 ## Backlog
 
 - [ ] **Frontend lint warnings cleanup.** 23 `react-hooks/exhaustive-deps` warnings remain after PR #149 (mostly missing-deps in useEffect). Either fix them or audit them for known-safe ones and add eslint-disable comments. Not blocking CI today.
 - [ ] **Audit `filterwarnings` ignores added in `wip(handoff): restore backend suite to green`.** Codex added narrow `ResourceWarning` filters for unclosed socket/transport/event-loop noise from pytest-asyncio teardown. Worth periodically reviewing whether those are still needed (e.g. when bumping pytest-asyncio) — if a real warning appears in those forms it would be silenced.
+- [ ] **Add `data-testid` attributes to e2e-critical interactive elements.** PR #152 fixed five Playwright tests by chasing UI-text changes (`Sessions` → `Session History`, `Account Settings` → `Account Management`, `/assistant` → `/pilot`, "Flow Sessions" tab, Resume button on session cards). Each was a one-line selector update, but every UI churn re-breaks them. Adding stable `data-testid` attributes on the targeted elements (page heading wrappers, tab nav, primary action buttons) and switching tests to `getByTestId` would make these immune to copy/route renames. Scope it small — start with `SessionHistoryPage` heading, the AI/Flow Sessions tab buttons, the per-session `Resume` button, and the command-palette FlowPilot option.
+- [ ] **Per-test transactional rollback in `test_db` fixture.** Bigger engineering than xdist (which we already shipped). Instead of `DROP SCHEMA public CASCADE` per test, wrap each test in a savepoint and rollback at teardown. ~30-40% additional speedup on top of xdist for test-DB-heavy tests. Real refactor; only worth it if the suite gets significantly larger or runs more frequently.
+- [ ] **Consider `pytest-testmon` for PR-time test selection.** Tracks which tests touched which source files and only re-runs affected ones. Best for small PRs touching ~few files. Adds cache-invalidation complexity; only worth it if the suite stays painfully long even after xdist.
+- [ ] **AssistantChatPage `currentChatRef` guard is a silent return** — `handleSend`, `handleTaskSubmit`, `selectChat`, `refreshFacts`, `refreshActiveFix`, and `refreshPreview` all bail with `if (currentChatRef.current !== sentForChatId) return` when stale. This is by design for chat switching, but it also silently masked the prefill-ref bug fixed in PR #153 — the user just saw "no AI response" with no log, no toast, no Sentry event. Either (a) log a `console.warn`/Sentry breadcrumb on the mismatch path so future drift is visible, or (b) split "expected stale" (chat switch) from "unexpected stale" (ref never updated) so only the latter alerts. Pair with an audit of every `currentChatRef.current = ...` assignment vs every `setActiveChatId(...)` call to make sure they're paired everywhere.

.gitea/workflows/ci.yml

@@ -17,10 +17,13 @@ jobs:
           POSTGRES_USER: postgres
           POSTGRES_PASSWORD: postgres
           POSTGRES_DB: resolutionflow_test
-        ports:
-          - 5432:5432
+        # No host port mapping. Tests connect to `postgres:5432` (the service
+        # container's docker-network DNS name), not `localhost:5432`. With
+        # multiple Gitea runners on the same homelab box, host-port mapping
+        # would race — two backend/e2e jobs both binding 0.0.0.0:5432 → the
+        # second fails with "port is already allocated".
         options: >-
-          --health-cmd pg_isready
+          --health-cmd "pg_isready -U postgres"
           --health-interval 10s
           --health-timeout 5s
           --health-retries 5
@@ -66,11 +69,15 @@ jobs:
         run: cd backend && python scripts/check_tenant_filters.py
 
       - name: Run tests with coverage
+        # `-n auto` parallelizes across all runner cores via pytest-xdist.
+        # conftest.py creates a per-worker DB (resolutionflow_test_gw0,
+        # resolutionflow_test_gw1, …) so the per-test DROP SCHEMA doesn't
+        # race across workers. Master/serial runs keep the base DB.
         # term-missing dropped — the custom "Display coverage summary" step
         # below parses coverage.json and prints the same info more concisely.
         # --maxfail=10 short-circuits on structural breakage so we don't burn
         # 25 minutes when a fixture explodes.
-        run: cd backend && python -m pytest --override-ini="addopts=" --maxfail=10 --cov=app --cov-report=json:coverage.json --cov-fail-under=50
+        run: cd backend && python -m pytest --override-ini="addopts=" -n auto --maxfail=10 --cov=app --cov-report=json:coverage.json --cov-fail-under=50
 
       - name: Display coverage summary
         if: always()
@@ -118,15 +125,14 @@ jobs:
       - name: Build
         run: cd frontend && NODE_OPTIONS="--max-old-space-size=4096" npm run build
 
-      - name: Upload build artifact
-        uses: actions/upload-artifact@v3
-        with:
-          name: frontend-dist
-          path: frontend/dist
-          retention-days: 1
+      # Build artifact intentionally NOT uploaded. The e2e job below builds
+      # its own frontend rather than downloading one from this job, so there
+      # is no need for the cross-job artifact handoff (which previously broke
+      # on actions/upload-artifact@v4 GHES support and forced a v3 pin).
+      # Decoupling also lets e2e start immediately rather than waiting for
+      # this job to finish — important on a multi-runner setup.
 
   e2e:
-    needs: [frontend]
     runs-on: ubuntu-latest
 
     services:
@@ -136,10 +142,13 @@ jobs:
           POSTGRES_USER: postgres
           POSTGRES_PASSWORD: postgres
           POSTGRES_DB: resolutionflow_test
-        ports:
-          - 5432:5432
+        # No host port mapping. Tests connect to `postgres:5432` (the service
+        # container's docker-network DNS name), not `localhost:5432`. With
+        # multiple Gitea runners on the same homelab box, host-port mapping
+        # would race — two backend/e2e jobs both binding 0.0.0.0:5432 → the
+        # second fails with "port is already allocated".
         options: >-
-          --health-cmd pg_isready
+          --health-cmd "pg_isready -U postgres"
           --health-interval 10s
           --health-timeout 5s
           --health-retries 5
@@ -152,6 +161,12 @@ jobs:
       PLAYWRIGHT_SECRET_KEY: ci-playwright-secret-key
       PLAYWRIGHT_TEST_EMAIL: teamadmin@resolutionflow.example.com
       PLAYWRIGHT_TEST_PASSWORD: TestPass123!
+      # AI-touching endpoints (POST /ai-sessions, /chat, /respond, etc.) are
+      # gated by `_require_ai_enabled()`, which returns 503 when no provider
+      # key is set. Tests that exercise those flows stub the AI calls in the
+      # browser via `page.route`, so the backend never actually contacts
+      # Anthropic — but the gate still has to pass. A stub value is enough.
+      ANTHROPIC_API_KEY: ci-stub-key-not-used-by-tests
 
     steps:
       - uses: actions/checkout@v4
@@ -178,11 +193,13 @@ jobs:
       - name: Install frontend dependencies
         run: cd frontend && npm ci
 
-      - name: Download frontend build
-        uses: actions/download-artifact@v3
-        with:
-          name: frontend-dist
-          path: frontend/dist
+      - name: Build frontend
+        # Building inline (instead of downloading an artifact from the
+        # frontend job) drops the cross-job dependency, so e2e can start
+        # immediately on a free runner. Adds ~1-2 min of build time, but
+        # eliminates the artifact-upload mechanism entirely (no more
+        # v3/v4 GHES headaches) and saves ~5 min of waiting.
+        run: cd frontend && NODE_OPTIONS="--max-old-space-size=4096" VITE_API_URL="${PLAYWRIGHT_API_ORIGIN}" npm run build
 
       - name: Install Playwright browser
         run: cd frontend && npx playwright install --with-deps chromium

backend/requirements-dev.txt

@@ -4,6 +4,7 @@
 # Testing — pytest-asyncio 0.24+ requires pytest>=8.2
 pytest==8.4.2
 pytest-asyncio==0.24.0
+pytest-xdist==3.6.1
 httpx>=0.27.0
 pytest-cov==5.0.0
 

backend/tests/conftest.py

@@ -35,11 +35,64 @@ settings.REQUIRE_INVITE_CODE = False
 # would silently nuke the dev database. Only DATABASE_TEST_URL is honored,
 # and the safety assertion below refuses to run against a DB whose name
 # doesn't contain "test".
-TEST_DATABASE_URL = os.environ.get(
+_BASE_TEST_DATABASE_URL = os.environ.get(
     "DATABASE_TEST_URL",
     "postgresql+asyncpg://postgres:postgres@localhost:5432/resolutionflow_test",
 )
 
+
+def _worker_db_url(base_url: str) -> str:
+    """Per-worker DB URL for pytest-xdist parallelization.
+
+    pytest-xdist sets PYTEST_XDIST_WORKER to 'gw0', 'gw1', ... per worker
+    process. Each worker needs its own database so the per-test
+    `DROP SCHEMA public CASCADE` doesn't race across workers. Master/serial
+    runs (no xdist) keep the base DB. The base DB is created by the postgres
+    service container; per-worker DBs are CREATE DATABASE-d on first import
+    by `_ensure_worker_db_exists` below.
+    """
+    worker = os.environ.get("PYTEST_XDIST_WORKER")
+    if not worker or worker == "master":
+        return base_url
+    head, tail = base_url.rsplit("/", 1)
+    db_name, _, query = tail.partition("?")
+    suffix = f"?{query}" if query else ""
+    return f"{head}/{db_name}_{worker}{suffix}"
+
+
+def _ensure_worker_db_exists(worker_url: str, base_url: str) -> None:
+    """Create the per-worker DB if it doesn't exist. Runs synchronously at
+    conftest import time (before any async test machinery), using psycopg2
+    against the postgres maintenance DB. No-op when not running under xdist.
+    """
+    if worker_url == base_url:
+        return
+    head, tail = worker_url.rsplit("/", 1)
+    worker_db = tail.partition("?")[0]
+    # Strip the +asyncpg dialect for sync psycopg2 + connect to 'postgres'.
+    sync_head = head.replace("+asyncpg", "")
+    admin_url = f"{sync_head}/postgres"
+    # Lazy import — psycopg2 is a transitive backend dep; not imported at
+    # module top to keep the conftest light when xdist isn't in use.
+    from sqlalchemy import create_engine
+    engine = create_engine(admin_url, isolation_level="AUTOCOMMIT")
+    try:
+        with engine.begin() as conn:
+            exists = conn.execute(
+                sa.text("SELECT 1 FROM pg_database WHERE datname = :n"),
+                {"n": worker_db},
+            ).scalar()
+            if not exists:
+                # Identifier interpolation is safe — worker_db is built from
+                # the trusted base URL + 'gw\d+' worker suffix.
+                conn.execute(sa.text(f'CREATE DATABASE "{worker_db}"'))
+    finally:
+        engine.dispose()
+
+
+TEST_DATABASE_URL = _worker_db_url(_BASE_TEST_DATABASE_URL)
+_ensure_worker_db_exists(TEST_DATABASE_URL, _BASE_TEST_DATABASE_URL)
+
 # Belt-and-suspenders: refuse to run tests against a DB whose name doesn't
 # contain "test". Parses the last path segment of the URL (everything after
 # the final '/', with query string stripped) so credentials / hosts that

frontend/e2e/assistant-chat-prefill.spec.ts

@@ -0,0 +1,111 @@
+import { expect, test } from '@playwright/test'
+
+/**
+ * Regression test for the prefill-handoff `currentChatRef` bug.
+ *
+ * Symptom: a chat session created via the dashboard prefill flow
+ * looked fine on the first AI turn, but submitting partial answers
+ * from the task lane silently dropped the AI's follow-up response.
+ * The user saw their answers in the chat, no assistant reply, no
+ * toast.
+ *
+ * Root cause: the prefill effect in `AssistantChatPage` set
+ * `activeChatId` without also updating `currentChatRef.current`, so
+ * the `currentChatRef.current !== sentForChatId` guard in
+ * `handleTaskSubmit` (and `handleSend`) tripped on every subsequent
+ * request and discarded the AI response.
+ *
+ * Strategy: drive the real prefill flow against the real backend, but
+ * intercept the `/chat` endpoint with `page.route` so we get
+ * deterministic question payloads on turn 1 and a deterministic
+ * follow-up on turn 2. The fix is what makes turn 2 visible.
+ */
+test.describe('AssistantChatPage — prefill handoff regression', () => {
+  test('AI follow-up renders after submitting partial task lane answers', async ({ page }) => {
+    let chatCallCount = 0
+
+    // Clear any persisted active-chat-id so the page does not auto-resume a
+    // stale session left behind by a sibling spec.
+    await page.addInitScript(() => {
+      try {
+        sessionStorage.removeItem('rf-active-chat-id')
+        sessionStorage.removeItem('rf-tasklane-meta')
+      } catch { /* ignore */ }
+    })
+
+    // Intercept only the chat endpoint. Session creation, listSessions,
+    // facts, suggested-fixes, etc. all hit the real backend so the page
+    // renders normally — only the LLM call is deterministic. The pattern
+    // matches `/ai-sessions/<uuid>/chat` and nothing nested beneath it.
+    await page.route(/\/api\/v1\/ai-sessions\/[^/]+\/chat$/, async (route) => {
+      if (route.request().method() !== 'POST') {
+        await route.fallback()
+        return
+      }
+      chatCallCount += 1
+      if (chatCallCount === 1) {
+        await route.fulfill({
+          status: 200,
+          contentType: 'application/json',
+          body: JSON.stringify({
+            content: 'Initial diagnostic plan. Please answer the questions in the task lane.',
+            suggested_flows: [],
+            fork: null,
+            actions: [],
+            questions: [
+              { text: 'Has the user recently changed their password?' },
+              { text: 'Is the lockout happening at a consistent time of day?' },
+            ],
+          }),
+        })
+        return
+      }
+      await route.fulfill({
+        status: 200,
+        contentType: 'application/json',
+        body: JSON.stringify({
+          content: 'Got it — based on your answer, here is what to check next.',
+          suggested_flows: [],
+          fork: null,
+          actions: [],
+          questions: [],
+        }),
+      })
+    })
+
+    // Drive the prefill flow exactly the way the dashboard does. The textarea
+    // is keyed by its placeholder copy on QuickStartPage.
+    await page.goto('/')
+    const prefillBox = page.getByPlaceholder(/Describe the issue/i)
+    await expect(prefillBox).toBeVisible({ timeout: 10_000 })
+    await prefillBox.fill('User locked out of AD weekly')
+    await prefillBox.press('Enter')
+
+    // After the prefill submits we land on /pilot and the first stubbed AI
+    // turn surfaces the task-lane question text.
+    await expect(page).toHaveURL(/\/pilot/)
+    await expect(
+      page.getByText('Has the user recently changed their password?'),
+    ).toBeVisible({ timeout: 15_000 })
+
+    // Answer the first question. UI flow: click "Answer" to open the
+    // textarea, type, click the inline "Answer" button to mark done.
+    await page.getByRole('button', { name: /^Answer$/ }).first().click()
+    await page.getByPlaceholder('Type your answer...').fill('No, password is months old')
+    await page.getByRole('button', { name: /^Answer$/ }).first().click()
+
+    // Submit the partial response. Pre-fix: the response was silently dropped
+    // here because `currentChatRef.current` still held the mount-time value.
+    await page.getByRole('button', { name: /Send 1 of 2 Responses/ }).click()
+
+    // Bug repro: the assistant message must render. Pre-fix this assertion
+    // fails because `handleTaskSubmit` early-returns at the
+    // `currentChatRef.current !== sentForChatId` guard.
+    await expect(
+      page.getByText('Got it — based on your answer, here is what to check next.'),
+    ).toBeVisible({ timeout: 15_000 })
+
+    // Both chat calls must have actually happened.
+    expect(chatCallCount).toBe(2)
+  })
+})

frontend/e2e/command-palette.spec.ts

@@ -88,6 +88,8 @@ test.describe('command palette smoke tests', () => {
 
     await flowpilotOption.click()
 
-    await expect(page).toHaveURL(/\/assistant/)
+    // Phase 1 of the FlowPilot migration renamed /assistant to /pilot.
+    // /assistant still 301-redirects to /pilot, so accept either landing URL.
+    await expect(page).toHaveURL(/\/(pilot|assistant)/)
   })
 })

frontend/e2e/history.spec.ts

@@ -24,13 +24,21 @@ test.describe('session history smoke tests', () => {
       await page.goto('/sessions')
 
       await expect(
-        page.getByRole('heading', { name: 'Sessions', exact: true }),
+        page.getByRole('heading', { name: 'Session History', exact: true }),
       ).toBeVisible()
 
+      // Default tab on /sessions is "AI Sessions"; flow sessions live behind
+      // the "Flow Sessions" tab and only that tab exposes ticket/client filters.
+      await page.getByRole('button', { name: 'Flow Sessions' }).click()
+
       await page.getByPlaceholder('Search by ticket number...').fill(ticketNumber)
       await page.getByPlaceholder('Search by client name...').fill(clientName)
 
-      const sessionCard = page.locator('.bg-card').filter({ hasText: ticketNumber }).filter({ hasText: clientName }).first()
+      const sessionCard = page
+        .getByTestId('flow-session-card')
+        .filter({ hasText: ticketNumber })
+        .filter({ hasText: clientName })
+        .first()
       await expect(sessionCard).toBeVisible()
       await expect(sessionCard.getByText(tree.name)).toBeVisible()
 

frontend/e2e/library-start.spec.ts

@@ -24,7 +24,7 @@ test.describe('flow library start-session smoke tests', () => {
       await page.getByPlaceholder('Search flows...').fill(tree.name)
       await page.getByRole('button', { name: 'Search', exact: true }).click()
 
-      const treeCard = page.locator('.bg-card').filter({ hasText: tree.name }).first()
+      const treeCard = page.getByTestId('tree-card').filter({ hasText: tree.name }).first()
       await expect(treeCard).toBeVisible()
       await treeCard.getByRole('button', { name: /^Start(?: Session)?$/ }).click()
 

frontend/e2e/library.spec.ts

@@ -20,7 +20,7 @@ test.describe('flow library smoke tests', () => {
       await page.getByPlaceholder('Search flows...').fill(tree.name)
       await page.getByRole('button', { name: 'Search', exact: true }).click()
 
-      await expect(page.getByText(tree.name)).toBeVisible()
+      await expect(page.getByTestId('tree-card').filter({ hasText: tree.name }).first()).toBeVisible()
     } finally {
       await disposeApiContext(api)
     }

frontend/e2e/navigation.spec.ts

@@ -14,7 +14,7 @@ test.describe('authenticated navigation smoke tests', () => {
     await page.goto('/sessions')
 
     await expect(
-      page.getByRole('heading', { name: 'Sessions', exact: true }),
+      page.getByRole('heading', { name: 'Session History', exact: true }),
     ).toBeVisible()
   })
 
@@ -30,7 +30,7 @@ test.describe('authenticated navigation smoke tests', () => {
     await page.goto('/account')
 
     await expect(
-      page.getByRole('heading', { name: 'Account Settings' }),
+      page.getByRole('heading', { name: 'Account Management' }),
     ).toBeVisible()
   })
 })

frontend/e2e/resume.spec.ts

@@ -18,9 +18,17 @@ test.describe('session resume smoke tests', () => {
     })
 
     try {
-      await page.goto('/trees')
+      // Resume flow moved off /trees onto the Flow Sessions tab of /sessions
+      // during the FlowPilot migration. The destination (/trees/:id/navigate)
+      // is unchanged — only the entry point shifted.
+      await page.goto('/sessions')
+      await expect(
+        page.getByRole('heading', { name: 'Session History', exact: true }),
+      ).toBeVisible()
+      await page.getByRole('button', { name: 'Flow Sessions' }).click()
+      // Active sub-tab is the default and surfaces in-progress sessions.
 
-      const resumeCard = page.locator('.bg-card').filter({ hasText: tree.name }).filter({ hasText: 'Resume' }).first()
+      const resumeCard = page.getByTestId('flow-session-card').filter({ hasText: tree.name }).first()
       await expect(resumeCard).toBeVisible()
       await resumeCard.getByRole('button', { name: 'Resume' }).first().click()
 

frontend/e2e/shares.spec.ts

@@ -31,7 +31,7 @@ test.describe('shared session management smoke tests', () => {
       ).toBeVisible()
       await expect(page.getByText(share.share_name || '')).toBeVisible()
 
-      const shareCard = page.locator('.bg-card').filter({ hasText: share.share_name || '' }).first()
+      const shareCard = page.getByTestId('share-card').filter({ hasText: share.share_name || '' }).first()
       await shareCard.getByRole('button', { name: 'Revoke' }).click()
 
       const confirmDialog = page.getByRole('dialog', { name: 'Revoke Share Link' })

frontend/src/components/library/TreeGridView.tsx

@@ -34,6 +34,8 @@ export function TreeGridView({
       {trees.map((tree) => (
         <div
           key={tree.id}
+          data-testid="tree-card"
+          data-tree-id={tree.id}
           className="relative bg-card border border-border rounded-2xl p-4 transition-all hover:-translate-y-0.5 hover:border-primary/30 hover:shadow-md sm:p-6"
         >
           <div className="mb-2 flex items-start justify-between gap-2">

frontend/src/components/library/TreeListView.tsx

@@ -33,6 +33,8 @@ export function TreeListView({
       {trees.map((tree) => (
         <div
           key={tree.id}
+          data-testid="tree-card"
+          data-tree-id={tree.id}
           className="flex items-center gap-4 bg-card border border-border rounded-2xl p-4 transition-all hover:border-primary/30 hover:shadow-xs"
         >
           {/* Left: Name and Description */}

frontend/src/pages/AssistantChatPage.tsx

@@ -255,6 +255,12 @@ export default function AssistantChatPage() {
         }
         setChats(prev => [chatItem, ...prev])
         setActiveChatId(session.session_id)
+        // Keep the in-flight guard ref in sync. Without this, currentChatRef
+        // stays at its mount-time value (often a stale id from sessionStorage
+        // or null), so subsequent handleSend / handleTaskSubmit calls bail at
+        // their `currentChatRef.current !== sentForChatId` check and the AI
+        // response is silently dropped.
+        currentChatRef.current = session.session_id
         setMessages([{ role: 'user', content: prefill }])
         setLoading(true)
 

frontend/src/pages/MySharesPage.tsx

@@ -161,7 +161,12 @@ export default function MySharesPage() {
             const isCopied = copiedId === share.id
 
             return (
-              <div key={share.id} className="bg-card border border-border rounded-xl p-5">
+              <div
+                key={share.id}
+                data-testid="share-card"
+                data-share-id={share.id}
+                className="bg-card border border-border rounded-xl p-5"
+              >
                 {/* Top row: badge + name */}
                 <div className="flex items-center gap-3 mb-3">
                   <span className="inline-flex items-center gap-1.5 text-xs rounded-full px-2 py-0.5 bg-accent text-muted-foreground">

frontend/src/pages/SessionHistoryPage.tsx

@@ -533,7 +533,11 @@ export default function SessionHistoryPage() {
                       )}
                       style={{ '--stagger-index': i } as React.CSSProperties}
                     >
-                      <div className="bg-card border border-border rounded-xl p-4 transition-all hover:border-[var(--color-border-hover)]">
+                      <div
+                        data-testid="flow-session-card"
+                        data-session-id={session.id}
+                        className="bg-card border border-border rounded-xl p-4 transition-all hover:border-[var(--color-border-hover)]"
+                      >
                         <div className="flex flex-col gap-3 sm:flex-row sm:items-start sm:justify-between">
                           <div className="flex-1">
                             <div className="flex flex-wrap items-center gap-2">