diff --git a/agents/README.md b/agents/README.md
index 3faa834ef..0fc6090a0 100644
--- a/agents/README.md
+++ b/agents/README.md
@@ -15,7 +15,7 @@ not fleet seats. Keep changes in `/mnt/storage/src/mosaic-stack`; do not modify
 | Dewey | Frontend design, UX, accessibility, and UI implementation | Pi, configured Mosaic model | `agents/dewey/launch.sh` |
 | Rocko | General development, investigation, testing, and review | T3 session (Codex, gpt-6.1-sol, thread d99326cc, lead decision 122; Claude thread b84bb264 retired 2026-10-12); launcher Claude Code, Sonnet model | `agents/rocko/launch.sh` |
 | Filbert | General development, investigation, testing, and review | T3 session (Codex, gpt-6.1-sol, thread 50342f6a, lead decision 122; Claude thread 9cb9731e retired 2026-10-12); launcher Pi, `openai-codex/gpt-6-astra:low`, not to be used under R26 until it changes (lead decision 69) | `agents/filbert/launch.sh` |
-| Researcher | Source-grounded technical research and evidence | Pi, configured Mosaic model | `agents/researcher/launch.sh` |
+| Researcher | Source-grounded technical research and evidence | T3 session (Claude, Sonnet 5.5, thread 9bac0794, rows 97 and 98, lead decision 124); Haiku rows (71, 80, 84, 88, later 96) run as subagents Sage dispatches with model haiku, effort high and a worktree (lead decision 129); Haiku thread c98aaf4a retired 2026-10-13 unused; launcher Pi, configured Mosaic model | `agents/researcher/launch.sh` |
 | Sage | Project lead: coordination, review, integration; earlier DYOR strategy records retained | T3 session (Claude Code); Pi launcher `zai/glm-5.3:high` retained | `agents/sage/launch.sh` |
 
 Each script supports `--check` and `--fresh`. Normal launches resume the agent's
diff --git a/docs/plans/2026-09-26_lead-decisions.md b/docs/plans/2026-09-26_lead-decisions.md
index 71c91dc86..4d47fd46d 100644
--- a/docs/plans/2026-09-26_lead-decisions.md
+++ b/docs/plans/2026-09-26_lead-decisions.md
@@ -2879,3 +2879,31 @@ which stay with him. Each item names who decided it and what happened.
     - Filbert's row 91 review says whether the session-live guard depends
       on anything else that assumes the role release means the process is
       gone.
+
+129. **Haiku rows run as subagents Sage dispatches (2026-10-13, Mos's
+    decision on Jason's Q130).** Jason restarted CLIProxyAPI and Mos got
+    HTTP 200 for claude-haiku-5-5 through :8317 at 00:03Z (Vikunja #242).
+    `t3_create_thread` can't set reasoning effort, but the Agent tool takes
+    model "haiku" and effort "high". Mos tested that at 00:06:06Z. Jason
+    asked for Haiku on High, which is why effort is set.
+    - Rows 71, 80, 84 and 88, and row 96 once its gate clears, run as
+      Agent subagents with model haiku, effort high and isolation worktree.
+      Each gets the full brief, its manifest and the decision-127 gate.
+      Reviews stay with the rows' named reviewers, Filbert among them
+      under decision 122, and Sage lands.
+    - The worktree's default base is origin/next. Each subagent detaches
+      to the stated refactor base first.
+    - The test slot still holds. A subagent builds without running tests
+      (phase A), then runs its suites and mutants only when Sage passes
+      the slot to it (phase B), and writes the packet to
+      `agents/researcher/work/queue-NN/rN/`. The queue owner stays
+      researcher, so the subagent starts its own row as researcher.
+      Sage posts review requests as jarvis and commits queue revisions.
+    - Rows 80 and 88 wait for row 94 to land. Row 80 shares
+      `packages/bus/README.md`, `broker.mjs` and `views.mjs` with it, and
+      row 88 shares `packages/webui/src/public/bus.js`,
+      `bus-browser.test.mjs` and the webui README. Decision 127 would
+      force a new round on whichever landed second.
+    - The Haiku T3 thread c98aaf4a ran at T3's default medium effort. It
+      got only a holding prompt and is retired unused. Sonnet Researcher
+      9bac0794 keeps rows 97 and 98.