fix(issue-57): review nits - a damaged audit sequence or an unresolvable log root never aborts the check

_check_disk_fallback converted transition_seq outside its try, so a
status.json whose sequence is not an integer would raise out of the
subscriber's main loop; such a status is now ignored with a warning, like a
stale one. delegate_started_recorded returned 1 before looking at the
registry's events.log when the audit-log root could not be resolved; it now
skips only the audit check.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

.agents/skills/lib.sh

@@ -2381,9 +2381,12 @@ delegate_started_recorded() {
   local reg_dir="${MAM_JOBS_DIR:-${WORKSPACE_ROOT:-$PWD}/.mam/jobs}"
   # Issue #57: the audit log is resolved by the one rule (next to the registry
   # unless DELEGATE_JOB_LOGS_DIR says otherwise); no bash copy of the default.
-  [ -n "$logs_dir" ] || logs_dir="$(delegate_logs_dir "$reg_dir")" || return 1
+  # If that resolution fails, only the audit check is skipped: the registry's
+  # events.log is still evidence.
+  [ -n "$logs_dir" ] || logs_dir="$(delegate_logs_dir "$reg_dir")" || logs_dir=""
   local ev_log="$reg_dir/$job_id.events.log"
-  local audit="$logs_dir/$job_id/events.ndjson"
+  local audit=""
+  [ -n "$logs_dir" ] && audit="$logs_dir/$job_id/events.ndjson"
   [ -f "$ev_log" ] || [ -f "$audit" ] || return 1
   EV_LOG="$ev_log" EV_AUDIT="$audit" EV_TOK="$tok" python3 -c '
 import hashlib, json, os, sys

.agents/skills/lib_py/delegation/job_subscriber.py

@@ -110,7 +110,11 @@ def _check_disk_fallback(
                 # ignores it. An unversioned record (never transitioned, or
                 # MAM_ENGINE_CAS=0) and a missing record keep the old rule.
                 aud_seq = logged.get("transition_seq")
-                if rec_seq is not None and (aud_seq is None or int(aud_seq) < int(rec_seq)):
+                try:
+                    stale = rec_seq is not None and (aud_seq is None or int(aud_seq) < int(rec_seq))
+                except (TypeError, ValueError):
+                    stale = True  # a damaged status.json proves nothing either
+                if stale:
                     logger.warning("disk fallback ignored audit status %r for %s: transition_seq %r is older "
                                    "than the record's %r", logged.get("status"), jid, aud_seq, rec_seq)
                 else:

tests/test_issue57_logs_follow_registry.py

@@ -369,6 +369,12 @@ def test_old_audit_terminal_does_not_complete_the_redispatched_attempt(moved, mo
     pending, terminal = {jid}, {}
     job_subscriber._check_disk_fallback(pending, str(reg), terminal, {jid: rec["dispatch_token"]}, {jid: 2})
     assert pending == {jid} and terminal == {}
+    # a damaged sequence is ignored too, and does not take the subscriber down
+    mqtt_common.job_log_path(jid, mqtt_common.STATUS_FILENAME, str(logs)).write_text(
+        json.dumps({"job_id": jid, "status": "completed", "transition_seq": "n/a"}))
+    pending, terminal = {jid}, {}
+    job_subscriber._check_disk_fallback(pending, str(reg), terminal, {jid: rec["dispatch_token"]}, {jid: 2})
+    assert pending == {jid} and terminal == {}
     # a newer audit-only terminal is still taken
     mqtt_common.update_logged_status(jid, "completed", logs_dir=str(logs), transition_seq=rec["transition_seq"] + 1)
     pending, terminal = {jid}, {}