diff --git a/formulas/run-supervisor.toml b/formulas/run-supervisor.toml
index ceaf340..9efeefe 100644
--- a/formulas/run-supervisor.toml
+++ b/formulas/run-supervisor.toml
@@ -137,15 +137,14 @@ For each finding from the health assessment, decide and execute an action.
 
 **P3 Stale PRs (CI done >20min, no push since):**
   Do NOT read dev-poll.sh, push branches, attempt merges, or investigate pipeline code.
-  Instead, file a vault item for the dev-agent to pick up:
-    Write $OPS_REPO_ROOT/vault/pending/stale-pr-${ISSUE_NUM}.md:
-      # Stale PR: ${PR_TITLE}
-      ## What
-      CI finished >20min ago but no git push has been made to the PR branch.
-      ## Why
-      P3 — Factory degraded: PRs should be pushed within 20min of CI completion.
-      ## Unblocks
-      - Factory health: dev-agent will push the branch and continue the workflow
+  Instead, nudge the dev-agent via tmux injection if a session is alive:
+    # Find the dev session for this issue
+    SESSION=$(tmux list-sessions -F '#{session_name}' 2>/dev/null | grep "dev-.*-${ISSUE_NUM}" | head -1)
+    if [ -n "$SESSION" ]; then
+      # Inject a nudge into the dev-agent session
+      tmux send-keys -t "$SESSION" "# [supervisor] PR stale >20min — CI finished, please push or update" Enter
+    fi
+  If no active tmux session exists, note it in the journal for the next dev-poll cycle.
   Do NOT file vault items for stale PRs unless they remain stale for >3 consecutive runs.
 
 ### Cannot auto-fix → file vault item
@@ -252,6 +251,7 @@ knowledge file in the ops repo:
 Knowledge files: memory.md, disk.md, ci.md, forge.md, dev-agent.md,
 review-agent.md, git.md.
 
-After writing the journal, the agent session completes automatically.
+After writing the journal, write the phase signal:
+  echo 'PHASE:done' > "$PHASE_FILE"
 """
 needs = ["report"]
diff --git a/supervisor/AGENTS.md b/supervisor/AGENTS.md
index db162d0..b2ea5e4 100644
--- a/supervisor/AGENTS.md
+++ b/supervisor/AGENTS.md
@@ -31,6 +31,8 @@ No action issues — the supervisor runs directly from cron like the planner and
   Claude evaluates all metrics and takes actions in a single interactive session
 - `$OPS_REPO_ROOT/knowledge/*.md` — Domain-specific remediation guides (memory,
   disk, CI, git, dev-agent, review-agent, forge)
+- `supervisor/supervisor-poll.sh` — Legacy bash orchestrator (superseded by
+  supervisor-run.sh + formula)
 
 **Alert priorities**: P0 (memory crisis), P1 (disk), P2 (factory stopped/stalled),
 P3 (degraded PRs, circular deps, stale deps), P4 (housekeeping).