diff --git a/docs/scratchpads/ri-050.md b/docs/scratchpads/ri-050.md
index 905daf60..65196476 100644
--- a/docs/scratchpads/ri-050.md
+++ b/docs/scratchpads/ri-050.md
@@ -220,3 +220,23 @@ and fargo. Protocol (both seats, every handoff):
   once green+reviewed: #1276 (docs) → #1277 (publish gate) → #1278/#1293/#1294 (code).
   After #1277 merges, watch the next push pipeline prove the verify gate live.
 - fargo's non-RI PRs (#1291/#1296/#1297/#1281) stay strictly his lane.
+
+## jarvis session 2026-08-17 (evening) — reviews, rebases, merge plan
+
+- Rebased #1276/#1277/#1278 onto 8199261c (heads 59e2c460 / 46784c8d / 4917df1f);
+  invariant tests 7/7 and forge 116/116 re-run green at new heads. #1270 touched
+  test-enumeration-exclusions.txt + package.json, NOT ci.yml — no semantic overlap with
+  #1277's ci.yml changes (checked, was a real concern).
+- Independent reviews recorded: #1293 APPROVED (review 173; macp 109/109; fail-closed paths
+  + aggregate state machine verified), #1294 APPROVED (review 174; prdy 20/20 + command
+  specs 9/9; single-writer + linkage persistence + labeled export + conflict-aware import
+  verified). Note: 19 unrelated mosaic suites fail on bare minimal install (known workspace
+  build-order artifact, documented by fargo) — not this change.
+- Measured: `next` has NO branch protection (API: only main listed). Cross-seat review
+  discipline is protocol-enforced, not Gitea-enforced. Flagged to fargo for Jason: direct
+  pushes to next trigger ungated publishes; protection is Jason's call (#1231 adjacent).
+- Merge order planned: #1276 (docs-only — no publish run) -> #1277 (first gated publish)
+  -> #1278 -> #1293 -> #1294. Sent fargo review requests with pinned head SHAs
+  (comms/20260818T011932Z__from-jarvis__a9c02b.md). Not merging #1293/#1294 before my three
+  clear fargo's review — order optimality beats speed; every pre-#1277 merge publishes ungated.
+- CI on the three rebased heads: pending at time of this entry.