diff --git a/.ai/HANDOFF.md b/.ai/HANDOFF.md
index c98093b2..4d0d7dc6 100644
--- a/.ai/HANDOFF.md
+++ b/.ai/HANDOFF.md
@@ -9,16 +9,22 @@
 subagent-driven-development on branch `feat/l1-ai-tree-builder-phase-2a`
 (branched from `main` @ `87236b5`; **not pushed**, `main` untouched).
 
-## ⚠️ Tooling note (read first — why this session stopped at Task 15)
-The harness's **Bash output channel AND the Read tool both became intermittently
-unreliable** late in the session — returning stale/cached output and, critically,
-*fabricating plausible content* (a Read of `router.tsx` invented an `L1CategoriesPage`
-import on line 13 that does not exist; confirmed absent by `grep -c` = 0 and the page
-file not existing). Work stopped at Task 16 because authoring router/settings edits
-requires first reading those files accurately, and that could no longer be trusted.
-**On resume: confirm the shell is reliable** (write a unique sentinel to a file, read
-it back; cross-check any Read against `grep`). Verified facts in this doc were
-sentinel-confirmed.
+## ⚠️ Tooling note (read first — why this session stopped at Task 16)
+The harness's **Bash output channel became intermittently unreliable** — returning
+stale/cached output (a Bash command that wrote `/tmp/perm.txt` instead returned a
+PRIOR command's `/tmp/vc.txt` content; a `cat` returned the wrong commit SHA). The
+Write/Edit channel stayed reliable; Read mostly reliable but occasionally served a
+stale temp file. Work stopped at Task 16 because wiring a new route/nav requires
+accurately reading `router.tsx` + `AccountSettingsPage.tsx` then editing them, and
+read-then-edit against stale reads is exactly what produced the broken Tasks 14–15
+earlier this session. **On resume: confirm the shell is reliable first** — write a
+unique sentinel to a file and read it back; cross-check any Read against a fresh
+`grep`; never commit without a sentinel-wrapped `tsc -b`/pytest verification whose
+unique sentinel you can see in the same output.
+
+Earlier-this-session gotcha that cost ~an hour: pytest `-p no:cov` conflicts with the
+`--cov` baked into `pytest.ini` addopts → pytest exits before running → `&& echo PASS`
+chains mislabel. Always use `--override-ini="addopts="`, never `-p no:cov`.
 
 Backend test invocation that works:
 `docker exec resolutionflow_backend pytest <path> --override-ini="addopts=" -q`
@@ -45,16 +51,23 @@ source_session_id + exactly-one CHECK + schema made optional); `l1_category_serv
 walked_path entry), `POST /l1/sessions/{id}/next-node`, `GET /l1/escalations`,
 `GET|PATCH /accounts/me/l1-categories`, `require_account_owner_or_admin` dep.
 
-**Frontend (Tasks 13–15) — committed, each tsc -b + eslint clean (sentinel-verified):**
-- `9c8f1e2` Task 13 — `types/l1.ts` (+ai_build, IntakeOutcome/Result, NearMiss, TreeNode,
+**Frontend (Tasks 13–15) — committed; whole-project `tsc -b` + eslint clean as of fix `2cc7c83`:**
+- `03e8748` Task 13 — `types/l1.ts` (+ai_build, IntakeOutcome/Result, NearMiss, TreeNode,
   NextNodeRequest/Result, L1Categories) + `api/l1.ts` (intake→IntakeResult; nextNode,
   escalations, getCategories, setCategories). nextNode body carries `node_text`.
-- `7b1e44a` Task 14 — `L1Dashboard.tsx` dispatch on outcome (matched/build→walker;
-  suggest→use-flow/build-new; out_of_scope→escalate-without-walk prompt).
-- `a3f0c21` Task 15 — `L1WalkTreeVariant.tsx` renders ai_build nodes via /next-node +
-  disclaimer banner (`bg-warning/10` — note: `*-dim` tokens are NOT defined as
-  `--color-*-dim`, use `/10` opacity modifiers); `L1WalkPage.tsx` routes ai_build →
-  tree variant. NOT browser-verified (couldn't launch chromium / shell degraded).
+- `df7150f` Task 14 + `f483196` Task 15 were FIRST committed BROKEN (I misread eslint
+  EXIT=0 as a clean tsc; both had TS2305 missing-export + props-contract errors).
+  **`2cc7c83` repairs both** — verified `npx tsc -b` EXIT=0 (whole project) + eslint EXIT=0:
+  - `L1Dashboard.tsx`: minimal outcome-dispatch edit on the REAL page (matched/build→walker;
+    suggest→use-flow/build-new; out_of_scope→escalate-without-walk).
+  - `L1WalkTreeVariant.tsx`: kept real props `{session,onSessionUpdate,onDone}` +
+    ResolveModal/EscalateModal + header + transcript sidebar; added ai_build branch that
+    walks nodes via /next-node (passes node_text), disclaimer banner (`bg-warning/10` —
+    NOTE: `*-dim` tokens are NOT `--color-*-dim`; use `/10` opacity), terminal→modals.
+    flow/proposal keep the Phase-1 synthetic path.
+  - `L1WalkPage.tsx` unchanged (already routes ai_build → tree variant).
+  NOT browser-verified (chromium can't launch here).
+- **Verified clean state: HEAD `2cc7c83`, working tree clean.**
 
 ## Resume point — Tasks 16–19