diff --git a/.github/workflows/agentic-token-audit.md b/.github/workflows/agentic-token-audit.md
index 5a4aef9..89c97d0 100644
--- a/.github/workflows/agentic-token-audit.md
+++ b/.github/workflows/agentic-token-audit.md
@@ -51,27 +51,46 @@ steps:
       set -euo pipefail
       mkdir -p /tmp/gh-aw/token-audit
 
-      # Download last 24 hours of agentic workflow logs as JSON
-      # Allow partial results — gh aw logs streams incrementally, so even if
-      # it hits an API rate limit partway through, the JSON written so far is
-      # still valid and should be processed by the agent.
-      LOGS_EXIT=0
-      gh aw logs \
-        --start-date -1d \
-        --json \
-        -c 100 \
-        > /tmp/gh-aw/token-audit/workflow-logs.json || LOGS_EXIT=$?
-
-      if [ -s /tmp/gh-aw/token-audit/workflow-logs.json ]; then
-        TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/workflow-logs.json)
-        echo "✅ Downloaded $TOTAL agentic workflow runs (last 24 hours)"
-        if [ "$LOGS_EXIT" -ne 0 ]; then
-          echo "⚠️ gh aw logs exited with code $LOGS_EXIT (partial results — likely API rate limit)"
+      # Download last 24 hours of agentic workflow logs as JSON, one workflow
+      # at a time. `gh aw logs` without a workflow filter scans repo-wide
+      # `gh run list` batches (newest-first, 250 runs each) and stops
+      # paginating as soon as one batch contains no processable agentic runs
+      # (skipped/cancelled runs are dropped before the empty-batch check —
+      # see github/gh-aw#38782). In a high-CI-volume repo a batch spans only
+      # a couple of hours, so the unfiltered call usually saw nothing but
+      # this run itself and reported an empty day. Workflow-scoped listing is
+      # unaffected by repo CI volume. Partial results are fine — each
+      # per-workflow file that was written successfully still gets merged.
+      PARTS_DIR=/tmp/gh-aw/token-audit/log-parts
+      mkdir -p "$PARTS_DIR"
+
+      for lock in .github/workflows/*.lock.yml; do
+        id=$(basename "$lock" .lock.yml)
+        PART_EXIT=0
+        gh aw logs "$id" \
+          --start-date -1d \
+          --json \
+          -c 100 \
+          > "$PARTS_DIR/$id.json" || PART_EXIT=$?
+        if [ -s "$PARTS_DIR/$id.json" ]; then
+          COUNT=$(jq '.runs | length' "$PARTS_DIR/$id.json" 2>/dev/null || echo 0)
+          echo "✅ $id: $COUNT runs (exit code $PART_EXIT)"
+        else
+          echo "⚠️ $id: no log data (exit code $PART_EXIT)"
+          rm -f "$PARTS_DIR/$id.json"
         fi
+      done
+
+      if ls "$PARTS_DIR"/*.json >/dev/null 2>&1; then
+        jq -s '{summary: {}, runs: (map(.runs // []) | add | unique_by(.run_id))}' \
+          "$PARTS_DIR"/*.json > /tmp/gh-aw/token-audit/workflow-logs.json
       else
-        echo "❌ No log data downloaded (exit code $LOGS_EXIT)"
+        echo "❌ No log data downloaded for any workflow"
         echo '{"runs":[],"summary":{}}' > /tmp/gh-aw/token-audit/workflow-logs.json
       fi
+
+      TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/workflow-logs.json)
+      echo "✅ Merged $TOTAL agentic workflow runs (last 24 hours)"
 timeout-minutes: 25
 ---
 
diff --git a/.github/workflows/agentic-token-optimizer.md b/.github/workflows/agentic-token-optimizer.md
index 6117f5e..df8d00d 100644
--- a/.github/workflows/agentic-token-optimizer.md
+++ b/.github/workflows/agentic-token-optimizer.md
@@ -40,24 +40,58 @@ steps:
 
       echo "📥 Downloading agentic workflow logs (last 7 days)..."
 
-      LOGS_EXIT=0
-      gh aw logs \
-        --start-date -7d \
-        --json \
-        -c 50 \
-        > /tmp/gh-aw/token-audit/all-runs.json || LOGS_EXIT=$?
-
-      if [ -s /tmp/gh-aw/token-audit/all-runs.json ]; then
-        TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/all-runs.json)
-        echo "✅ Downloaded $TOTAL agentic workflow runs (last 7 days)"
-        if [ "$LOGS_EXIT" -ne 0 ]; then
-          echo "⚠️ gh aw logs exited with code $LOGS_EXIT (partial results — likely API rate limit)"
+      # Fetch logs one workflow at a time. `gh aw logs` without a workflow
+      # filter scans repo-wide `gh run list` batches (newest-first, 250 runs
+      # each) and stops paginating as soon as one batch contains no
+      # processable agentic runs (skipped/cancelled runs are dropped before
+      # the empty-batch check — see github/gh-aw#38782). In a high-CI-volume
+      # repo a batch spans only a couple of hours, so the unfiltered call
+      # truncates the candidate pool to whatever ran most recently.
+      # Workflow-scoped listing is unaffected by repo CI volume. Partial
+      # results are fine — each per-workflow file that was written
+      # successfully still gets merged.
+      PARTS_DIR=/tmp/gh-aw/token-audit/log-parts
+      mkdir -p "$PARTS_DIR"
+
+      for lock in .github/workflows/*.lock.yml; do
+        id=$(basename "$lock" .lock.yml)
+        PART_EXIT=0
+        gh aw logs "$id" \
+          --start-date -7d \
+          --json \
+          -c 50 \
+          > "$PARTS_DIR/$id.json" || PART_EXIT=$?
+        if [ -s "$PARTS_DIR/$id.json" ]; then
+          COUNT=$(jq '.runs | length' "$PARTS_DIR/$id.json" 2>/dev/null || echo 0)
+          echo "✅ $id: $COUNT runs (exit code $PART_EXIT)"
+        else
+          echo "⚠️ $id: no log data (exit code $PART_EXIT)"
+          rm -f "$PARTS_DIR/$id.json"
         fi
+      done
+
+      if ls "$PARTS_DIR"/*.json >/dev/null 2>&1; then
+        jq -s '{summary: {}, runs: (map(.runs // []) | add | unique_by(.run_id))}' \
+          "$PARTS_DIR"/*.json > /tmp/gh-aw/token-audit/all-runs.json
       else
-        echo "❌ No log data downloaded (exit code $LOGS_EXIT)"
+        echo "❌ No log data downloaded for any workflow"
         echo '{"runs":[],"summary":{}}' > /tmp/gh-aw/token-audit/all-runs.json
       fi
 
+      TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/all-runs.json)
+      echo "✅ Merged $TOTAL agentic workflow runs (last 7 days)"
+
+      # Exclude the AIC monitoring family (this optimizer + its sibling audit) from the
+      # candidate pool so the optimizer never selects its own meta-monitoring workflows.
+      # The in-prompt "Token in the name" guard misses these: their display names are
+      # "...AIC Usage Optimizer/Audit" (no "Token"), so match on workflow id/name here.
+      jq '.runs |= map(select(
+            (((.workflow_path // "") | test("agentic-token-(optimizer|audit)"))
+             or ((.workflow_name // "") | test("AIC Usage (Optimizer|Audit)"))) | not
+          ))' /tmp/gh-aw/token-audit/all-runs.json > /tmp/gh-aw/token-audit/all-runs.filtered.json \
+        && mv /tmp/gh-aw/token-audit/all-runs.filtered.json /tmp/gh-aw/token-audit/all-runs.json
+      echo "🚫 Excluded AIC monitoring family — $(jq '.runs | length' /tmp/gh-aw/token-audit/all-runs.json) runs remain in candidate pool"
+
   - name: Aggregate top workflows by AIC usage
     run: |
       set -euo pipefail
@@ -161,7 +195,7 @@ Treat missing numeric fields (`aic`, `token_usage`, `turns`, `action_minutes`) a
 
 - Start from `top-workflows.json`.
 - Exclude workflows optimized in the last 14 days (use `optimization-log.json`).
-- Exclude workflows with "Token" in the name to avoid self-targeting.
+- Exclude the AIC monitoring family — the `agentic-token-optimizer` and `agentic-token-audit` workflows (display names "Agentic Workflow AIC Usage Optimizer" / "Daily Agentic Workflow AIC Usage Audit") — to avoid self-targeting. These are also pre-filtered from `all-runs.json`/`top-workflows.json`, but never select them even if a stale snapshot still lists them.
 - Choose the highest AI-credit-spend workflow that remains.
 - If no snapshot/history exists, derive candidates directly from `all-runs.json`.
 
diff --git a/workflows/agentic-token-audit.md b/workflows/agentic-token-audit.md
index 5a4aef9..89c97d0 100644
--- a/workflows/agentic-token-audit.md
+++ b/workflows/agentic-token-audit.md
@@ -51,27 +51,46 @@ steps:
       set -euo pipefail
       mkdir -p /tmp/gh-aw/token-audit
 
-      # Download last 24 hours of agentic workflow logs as JSON
-      # Allow partial results — gh aw logs streams incrementally, so even if
-      # it hits an API rate limit partway through, the JSON written so far is
-      # still valid and should be processed by the agent.
-      LOGS_EXIT=0
-      gh aw logs \
-        --start-date -1d \
-        --json \
-        -c 100 \
-        > /tmp/gh-aw/token-audit/workflow-logs.json || LOGS_EXIT=$?
-
-      if [ -s /tmp/gh-aw/token-audit/workflow-logs.json ]; then
-        TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/workflow-logs.json)
-        echo "✅ Downloaded $TOTAL agentic workflow runs (last 24 hours)"
-        if [ "$LOGS_EXIT" -ne 0 ]; then
-          echo "⚠️ gh aw logs exited with code $LOGS_EXIT (partial results — likely API rate limit)"
+      # Download last 24 hours of agentic workflow logs as JSON, one workflow
+      # at a time. `gh aw logs` without a workflow filter scans repo-wide
+      # `gh run list` batches (newest-first, 250 runs each) and stops
+      # paginating as soon as one batch contains no processable agentic runs
+      # (skipped/cancelled runs are dropped before the empty-batch check —
+      # see github/gh-aw#38782). In a high-CI-volume repo a batch spans only
+      # a couple of hours, so the unfiltered call usually saw nothing but
+      # this run itself and reported an empty day. Workflow-scoped listing is
+      # unaffected by repo CI volume. Partial results are fine — each
+      # per-workflow file that was written successfully still gets merged.
+      PARTS_DIR=/tmp/gh-aw/token-audit/log-parts
+      mkdir -p "$PARTS_DIR"
+
+      for lock in .github/workflows/*.lock.yml; do
+        id=$(basename "$lock" .lock.yml)
+        PART_EXIT=0
+        gh aw logs "$id" \
+          --start-date -1d \
+          --json \
+          -c 100 \
+          > "$PARTS_DIR/$id.json" || PART_EXIT=$?
+        if [ -s "$PARTS_DIR/$id.json" ]; then
+          COUNT=$(jq '.runs | length' "$PARTS_DIR/$id.json" 2>/dev/null || echo 0)
+          echo "✅ $id: $COUNT runs (exit code $PART_EXIT)"
+        else
+          echo "⚠️ $id: no log data (exit code $PART_EXIT)"
+          rm -f "$PARTS_DIR/$id.json"
         fi
+      done
+
+      if ls "$PARTS_DIR"/*.json >/dev/null 2>&1; then
+        jq -s '{summary: {}, runs: (map(.runs // []) | add | unique_by(.run_id))}' \
+          "$PARTS_DIR"/*.json > /tmp/gh-aw/token-audit/workflow-logs.json
       else
-        echo "❌ No log data downloaded (exit code $LOGS_EXIT)"
+        echo "❌ No log data downloaded for any workflow"
         echo '{"runs":[],"summary":{}}' > /tmp/gh-aw/token-audit/workflow-logs.json
       fi
+
+      TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/workflow-logs.json)
+      echo "✅ Merged $TOTAL agentic workflow runs (last 24 hours)"
 timeout-minutes: 25
 ---
 
diff --git a/workflows/agentic-token-optimizer.md b/workflows/agentic-token-optimizer.md
index 6117f5e..df8d00d 100644
--- a/workflows/agentic-token-optimizer.md
+++ b/workflows/agentic-token-optimizer.md
@@ -40,24 +40,58 @@ steps:
 
       echo "📥 Downloading agentic workflow logs (last 7 days)..."
 
-      LOGS_EXIT=0
-      gh aw logs \
-        --start-date -7d \
-        --json \
-        -c 50 \
-        > /tmp/gh-aw/token-audit/all-runs.json || LOGS_EXIT=$?
-
-      if [ -s /tmp/gh-aw/token-audit/all-runs.json ]; then
-        TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/all-runs.json)
-        echo "✅ Downloaded $TOTAL agentic workflow runs (last 7 days)"
-        if [ "$LOGS_EXIT" -ne 0 ]; then
-          echo "⚠️ gh aw logs exited with code $LOGS_EXIT (partial results — likely API rate limit)"
+      # Fetch logs one workflow at a time. `gh aw logs` without a workflow
+      # filter scans repo-wide `gh run list` batches (newest-first, 250 runs
+      # each) and stops paginating as soon as one batch contains no
+      # processable agentic runs (skipped/cancelled runs are dropped before
+      # the empty-batch check — see github/gh-aw#38782). In a high-CI-volume
+      # repo a batch spans only a couple of hours, so the unfiltered call
+      # truncates the candidate pool to whatever ran most recently.
+      # Workflow-scoped listing is unaffected by repo CI volume. Partial
+      # results are fine — each per-workflow file that was written
+      # successfully still gets merged.
+      PARTS_DIR=/tmp/gh-aw/token-audit/log-parts
+      mkdir -p "$PARTS_DIR"
+
+      for lock in .github/workflows/*.lock.yml; do
+        id=$(basename "$lock" .lock.yml)
+        PART_EXIT=0
+        gh aw logs "$id" \
+          --start-date -7d \
+          --json \
+          -c 50 \
+          > "$PARTS_DIR/$id.json" || PART_EXIT=$?
+        if [ -s "$PARTS_DIR/$id.json" ]; then
+          COUNT=$(jq '.runs | length' "$PARTS_DIR/$id.json" 2>/dev/null || echo 0)
+          echo "✅ $id: $COUNT runs (exit code $PART_EXIT)"
+        else
+          echo "⚠️ $id: no log data (exit code $PART_EXIT)"
+          rm -f "$PARTS_DIR/$id.json"
         fi
+      done
+
+      if ls "$PARTS_DIR"/*.json >/dev/null 2>&1; then
+        jq -s '{summary: {}, runs: (map(.runs // []) | add | unique_by(.run_id))}' \
+          "$PARTS_DIR"/*.json > /tmp/gh-aw/token-audit/all-runs.json
       else
-        echo "❌ No log data downloaded (exit code $LOGS_EXIT)"
+        echo "❌ No log data downloaded for any workflow"
         echo '{"runs":[],"summary":{}}' > /tmp/gh-aw/token-audit/all-runs.json
       fi
 
+      TOTAL=$(jq '.runs | length' /tmp/gh-aw/token-audit/all-runs.json)
+      echo "✅ Merged $TOTAL agentic workflow runs (last 7 days)"
+
+      # Exclude the AIC monitoring family (this optimizer + its sibling audit) from the
+      # candidate pool so the optimizer never selects its own meta-monitoring workflows.
+      # The in-prompt "Token in the name" guard misses these: their display names are
+      # "...AIC Usage Optimizer/Audit" (no "Token"), so match on workflow id/name here.
+      jq '.runs |= map(select(
+            (((.workflow_path // "") | test("agentic-token-(optimizer|audit)"))
+             or ((.workflow_name // "") | test("AIC Usage (Optimizer|Audit)"))) | not
+          ))' /tmp/gh-aw/token-audit/all-runs.json > /tmp/gh-aw/token-audit/all-runs.filtered.json \
+        && mv /tmp/gh-aw/token-audit/all-runs.filtered.json /tmp/gh-aw/token-audit/all-runs.json
+      echo "🚫 Excluded AIC monitoring family — $(jq '.runs | length' /tmp/gh-aw/token-audit/all-runs.json) runs remain in candidate pool"
+
   - name: Aggregate top workflows by AIC usage
     run: |
       set -euo pipefail
@@ -161,7 +195,7 @@ Treat missing numeric fields (`aic`, `token_usage`, `turns`, `action_minutes`) a
 
 - Start from `top-workflows.json`.
 - Exclude workflows optimized in the last 14 days (use `optimization-log.json`).
-- Exclude workflows with "Token" in the name to avoid self-targeting.
+- Exclude the AIC monitoring family — the `agentic-token-optimizer` and `agentic-token-audit` workflows (display names "Agentic Workflow AIC Usage Optimizer" / "Daily Agentic Workflow AIC Usage Audit") — to avoid self-targeting. These are also pre-filtered from `all-runs.json`/`top-workflows.json`, but never select them even if a stale snapshot still lists them.
 - Choose the highest AI-credit-spend workflow that remains.
 - If no snapshot/history exists, derive candidates directly from `all-runs.json`.