fix: recover agent state on timeout instead of losing progress

When Claude timed out (exit 124), `set -euo pipefail` in agent-loop.sh
prevented sync_state from running, so all work from that iteration was
lost. Additionally, entrypoint.sh treated timeout as fatal and broke
the loop, preventing any retry.

Now run_claude captures the timeout exit code so sync_state always runs
to preserve progress, and entrypoint.sh continues the loop on timeout
so the next iteration can pick up where the previous one left off.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: LeeCampbell

autonomous/agent-loop.sh

@@ -57,10 +57,12 @@ sync_state() {
     fi
 }
 
+CLAUDE_RC=0
+
 run_claude() {
     local prompt="$1"
     timeout "$CLAUDE_TIMEOUT" claude --dangerously-skip-permissions --print \
-        --output-format stream-json --verbose "$prompt"
+        --output-format stream-json --verbose "$prompt" || CLAUDE_RC=$?
 }
 
 load_prompt() {
@@ -288,3 +290,5 @@ EOF
         fi
         ;;
 esac
+
+exit $CLAUDE_RC

autonomous/entrypoint.sh

@@ -27,7 +27,7 @@ echo "Building..."
 dotnet build --no-restore
 
 # ── Run the agent ──
-MAX_ITERATIONS="${MAX_ITERATIONS:-10}"
+MAX_ITERATIONS="${MAX_ITERATIONS:-30}"
 COOLDOWN="${COOLDOWN_SECONDS:-30}"
 export CLAUDE_TIMEOUT="${TIMEOUT_SECONDS:-1800}"
 
@@ -39,9 +39,11 @@ for i in $(seq 1 "$MAX_ITERATIONS"); do
     EXIT_CODE=0
     bash /usr/local/bin/agent-loop.sh || EXIT_CODE=$?
 
-    if [ "$EXIT_CODE" -ne 0 ]; then
+    if [ "$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..."
     fi
 
     # Done?