decisions: 96, Q97 yes, next-target plan confirmed as written

Co-Authored-By: Claude Opus 5.5 <[email protected]>

docs/SESSIONS.md

@@ -673,3 +673,4 @@ are never rewritten or removed; corrections are new entries.
 - 2026-10-11T22:27Z, Sage (lead): next-target plan sent to Mos for discussion; nothing moves. It covers Mos in the stack (rows M1-M3), iterating inside the deployment (D1-D4), and install paths with containers recommended and local supported (I1-I6; the pi worker image is the only container today). The open question to Mos is whether the Researcher starts the read-only I1 survey now. Darkwing's row 58 round 1 is in review, noted.
 - 2026-10-11T22:33Z, Sage (lead): Mos mos-0059 asked for two things. (1) The Researcher (T3 9bac0794) has the read-only I1 container survey, due 2026-10-12T18:00Z, written to agents/researcher/work/i1-container-survey/REPORT.md. (2) I drafted M1, a role for Jason's main agent, at agents/sage/work/next-target/M1-draft.md. It's a draft only, not in docs/plans or the queue, and waits on Q97 and Jason's approval of the role file. The I1 draft waits on the survey.
 - 2026-10-11T22:29Z, Sage (lead): correction. The line above stamped 22:33Z was estimated, not taken from the clock. The real time was about 22:29Z (commit a3d17005).
+- 2026-10-11T22:31Z, Sage (lead): Mos mos-0060, Q97 yes. Lead decision 96 records the next-target plan as written: S7, then M1-M2, then D1-D2 with I1, then I2-I6, with I4 before M3. M1 and I1 move to docs/plans and the queue after the cutover rows.

docs/plans/2026-09-26_lead-decisions.md

@@ -2073,3 +2073,31 @@ which stay with him. Each item names who decided it and what happened.
       the likeliest way the date slips. The cost is that the bus trail
       records a web send exactly as a CLI send, `human-cli`, until the
       label row.
+
+96. **Q97 yes: the next-target plan (2026-10-11).** Jason, relayed by
+    Mos (mos-0060): "Q97: yes". Mos reads it as the plan Sage sent
+    confirmed as written, not the faster option of M1 plus row 76.
+    - Three tracks:
+      - Mos in the stack (M1 a role for the main agent, M2 a thread
+        view, M3 a trial beside T3 Mos).
+      - Iterating inside the deployment (D1 a stack release and
+        rollback, D2 staging, D3 managed coders and reviewers, D4 CI).
+      - Install paths, containerized recommended and local supported
+        (I1 the design, I2 to I6 the images, the human proof inside
+        containers and the install mode).
+    - Order from 2026-10-13: S7, then M1-M2, then D1-D2 alongside I1,
+      then I2-I6. I4, a Console with signed-in human authority, lands
+      before M3 retires T3 Mos. The M3 trial is about 10-17. The
+      install track ends about 10-24, 10-31 at worst.
+    - Briefs and rows: the M1 and I1 drafts under
+      `agents/sage/work/next-target/` move to `docs/plans/` and the
+      queue after the cutover rows (74-76) and ops-01's route. Those
+      keep priority. The Researcher's read-only I1 survey goes on (due
+      2026-10-12T18:00Z).
+    - Still Jason's: approving the M1 role file (a policy change),
+      ratifying I1, and when T3 Mos retires. Each goes to Mos as a
+      short item when it's ready.
+    - The tradeoff: Mos stays a T3 thread for about a week longer than
+      the faster option. In exchange, the in-stack Mos never runs with
+      its wider authority under Jason's user, with PID-only isolation,
+      for more than the trial.