From cf13db6869a38531dc6d05ba1f7614900c012c34 Mon Sep 17 00:00:00 2001
From: Lee Campbell <lee.ryan.campbell@gmail.com>
Date: Tue, 24 Mar 2026 07:37:45 +0800
Subject: [PATCH] fix: ensure agent loop continues after Claude timeout or API
 errors

The agent loop would stop iterating when Claude exited with a non-zero,
non-124 exit code. This prevented the state machine from advancing to
the create-pr state after a successful execute-tasks phase where Claude
timed out but sync_state had already committed and pushed the work.

- Remove global CLAUDE_RC; run_claude now returns its exit code directly
- Add || true to all run_claude call sites so sync_state always runs
- Log Claude exit codes as warnings rather than swallowing them
- entrypoint.sh no longer breaks the loop on non-zero exit codes,
  letting the state machine advance to the next phase

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 autonomous/agent-loop.sh | 23 ++++++++++++-----------
 autonomous/entrypoint.sh |  7 ++-----
 2 files changed, 14 insertions(+), 16 deletions(-)

diff --git a/autonomous/agent-loop.sh b/autonomous/agent-loop.sh
index befc8dd..a5207da 100644
--- a/autonomous/agent-loop.sh
+++ b/autonomous/agent-loop.sh
@@ -57,12 +57,15 @@ sync_state() {
     fi
 }
 
-CLAUDE_RC=0
-
 run_claude() {
     local prompt="$1"
+    local rc=0
     timeout "$CLAUDE_TIMEOUT" claude --dangerously-skip-permissions --print \
-        --output-format stream-json --verbose "$prompt" || CLAUDE_RC=$?
+        --output-format stream-json --verbose "$prompt" || rc=$?
+    if [ "$rc" -ne 0 ]; then
+        echo "WARNING: claude exited with code $rc" >&2
+    fi
+    return "$rc"
 }
 
 load_prompt() {
@@ -197,25 +200,25 @@ EOF
 
     execute-tasks)
         ISSUE_NUM=$(get_issue_num)
-        run_claude "$(load_prompt execute-tasks)"
+        run_claude "$(load_prompt execute-tasks)" || true
         sync_state "feat(#${ISSUE_NUM}): implement tasks"
         ;;
 
     create-tasks)
         ISSUE_NUM=$(get_issue_num)
-        run_claude "$(load_prompt create-tasks)"
+        run_claude "$(load_prompt create-tasks)" || true
         sync_state "plan(#${ISSUE_NUM}): create task breakdown"
         ;;
 
     apply-review)
         ISSUE_NUM=$(get_issue_num)
-        run_claude "$(load_prompt apply-review)"
+        run_claude "$(load_prompt apply-review)" || true
         sync_state "plan(#${ISSUE_NUM}): apply brief review feedback"
         ;;
 
     review-brief)
         ISSUE_NUM=$(get_issue_num)
-        run_claude "$(load_prompt review-brief)"
+        run_claude "$(load_prompt review-brief)" || true
         sync_state "plan(#${ISSUE_NUM}): review brief"
         ;;
 
@@ -269,7 +272,7 @@ EOF
                 PROMPT=$(load_prompt pick-issue)
                 PROMPT="${PROMPT//\{\{ISSUE_BODY\}\}/$ISSUE_BODY}"
 
-                run_claude "$PROMPT"
+                run_claude "$PROMPT" || true
                 sync_state "plan(#${ISSUE_NUM}): initial brief from issue"
             fi
         else
@@ -285,10 +288,8 @@ EOF
             PROMPT=$(load_prompt pick-issue)
             PROMPT="${PROMPT//\{\{ISSUE_BODY\}\}/$ISSUE_BODY}"
 
-            run_claude "$PROMPT"
+            run_claude "$PROMPT" || true
             sync_state "plan(#${ISSUE_NUM}): initial brief from issue"
         fi
         ;;
 esac
-
-exit $CLAUDE_RC
diff --git a/autonomous/entrypoint.sh b/autonomous/entrypoint.sh
index 0c0be31..643c2da 100644
--- a/autonomous/entrypoint.sh
+++ b/autonomous/entrypoint.sh
@@ -39,11 +39,8 @@ for i in $(seq 1 "$MAX_ITERATIONS"); do
     EXIT_CODE=0
     bash /usr/local/bin/agent-loop.sh || EXIT_CODE=$?
 
-    if [ "$EXIT_CODE" -ne 0 ] && [ "$EXIT_CODE" -ne 124 ]; then
-        echo "Iteration $i failed with exit code $EXIT_CODE"
-        break
-    elif [ "$EXIT_CODE" -eq 124 ]; then
-        echo "Iteration $i timed out (exit code 124), continuing..."
+    if [ "$EXIT_CODE" -ne 0 ]; then
+        echo "Iteration $i exited with code $EXIT_CODE, continuing to next state..."
     fi
 
     # Done?