Skip to main content

AgentXchain v2.155.70

v2.155.70 closes BUG-114, found during the strict Tusq DOGFOOD-100 reverify after v2.155.69 recovered the retained Claude Node runtime ghost blocker. The recovered QA turn correctly landed on a typed dispatch:claude_auth_failed blocker, but the continuous loop still required operator step --resume after credentials were available to the resumed process. That violated the full-auto proof rule.

Bug Fixes​

  • BUG-114: refreshed Claude auth recovery. Continuous startup now auto-reissues retained Claude auth dispatch blockers once the current process has Claude auth env available.
  • Fail-closed boundary. Recovery only runs for retained Claude local_cli turns whose dispatch logs contain auth-failure markers. Non-Claude runtimes, missing logs, missing credentials, and logs without auth evidence remain blocked.
  • Audit evidence. Auto-recovery emits auto_retried_ghost with failure_type: "claude_auth_failed" and recovery_class: "claude_auth_refreshed" so dogfood proof can distinguish auth-refresh recovery from generic ghost retry.

Evidence​

  • node --test --test-name-pattern='BUG-114|BUG-113|BUG-112|BUG-111' test/continuous-run.test.js -> 6 tests / 2 suites / 0 failures
  • node --test test/run-events.test.js -> 12 tests / 1 suite / 0 failures
  • node --test test/continuous-run.test.js -> 84 tests / 13 suites / 0 failures
  • node --test test/local-cli-adapter.test.js test/human-roadmap-open-blockers-content.test.js test/run-events.test.js -> 59 tests / 7 suites / 0 failures
  • npm test -- --test-timeout=60000 -> 7333 tests / 1485 suites / 0 failures / 3 skipped

Tester Re-Run Contract​

Verify: npx --yes -p [email protected] -c "agentxchain --version"

For BUG-114: resume tusq.dev from continuous session cont-7dc5b5df, run run_73ffb608f7c8a510, and retained QA turn turn_aa521bedd41f1655 using the shipped package with Claude credentials available to the process. Startup must reissue the retained auth blocker without agentxchain step --resume, agentxchain unblock, manual state edits, staging JSON edits, or operator-side accept-turn.

If the recovered QA turn accepts, append counter value 98 to Tusq turn-counter.jsonl within the DOGFOOD-100 window.

For BUG-113 carry-forward: retained Claude Node runtime ghost blockers must still reissue through the compatible Node wrapper when old-Node signatures appear in dispatch logs.

For BUG-112 carry-forward: retained Claude provider Request timed out failures must still auto-reissue via auto_retried_productive_timeout only when explicit timeout markers exist.

For BUG-111 carry-forward: retained Claude 401/authentication failures must still reclassify to dispatch:claude_auth_failed before timeout or ghost recovery.

For BUG-110 carry-forward: fresh Claude local CLI auth failures must still block once as dispatch infrastructure instead of consuming turn retries.

For BUG-109 carry-forward: supplemental auto-checkpoint recovery must still run automatically under --auto-checkpoint.

For BUG-108 carry-forward: terminal blocked steps must surface once and must not re-enter paused-active recovery loops.

For BUG-107 carry-forward: paused active continuous sessions must still preserve the existing session ID and recover only when the governed run is active, unblocked, and dispatchable.

Carry-forward tester checks for shipped-package closure proof:

  • BUG-54 startup watchdog evidence should quote startup_latency_ms, elapsed_since_spawn_ms, and runtimes.<id>.startup_watchdog_ms.
  • BUG-55/87 verification-output evidence should quote undeclared_verification_outputs, verification.produced_files, and git status --short after accept-turn followed by checkpoint-turn. Clean tree means BUG-55 is fixed for your reproduction.
  • BUG-52/53 continuous recovery evidence should quote phase_entered, reconciled_before_dispatch, and session_continuation.