fix(sync): resolve AttributeError and make Telegram config optional

fix(sync): disable verbose file listing to speed up startup

feat(logging): enhance sync and app logging for debugging

fix: ensure telegram config persists in sync logic

Redirect sync logs to workspace/sync.log for debugging

Add credential reset logic to sync_hf.py

Implement robust credential reset via marker file

Fix WhatsApp guardian to handle 440 Conflict and remove legacy save script

Update default model to glm-4.5-flash

Enable log syncing for debug loop

Update Zhipu API config to use user-provided credentials and correct model ID

Fix Zhipu config: use 'openai-completions' api type

Fix Zhipu config: use 'openai' api type instead of 'openai-chat'

Capture application logs to workspace/startup.log for debugging

Try run dist/entry.js gateway

Fix entry point: run dist/entry.js instead of openclaw.mjs

Fix race condition: serialize startup of guardians after restore

feat: sync-on-change optimization & update config to GLM-4.7-FlashX

fix: use snapshot_download for folder restore

fix: debug logging for download failures and file listing

fix: define missing MAX_SESSION_AGE constant

fix: exclude models.json with secrets from agent sync

fix: restore missing __init__ method causing AttributeError

fix: restore missing OpenClawHFSync class definition

fix: reduce sync interval to 30s and add verbose credentials logging

fix: enable unbuffered output for sync script

fix: patch openclaw.json to remove stale plugins config and enforce model

feat: unified hf sync

feat: Migrate dataset backups to gzipped tar archives and ensure backward compatibility during restore and rotation.

feat: enhance persistence (logs/extensions) and restore GLM-4-Plus

chore: force glm-4-flash in entrypoint patch for verification

chore: temp switch to glm-4-flash for verification

fix: enforce model config update and trigger immediate persistence on login

fix: resolve NameError in restore script and cleanup imports

fix: update model to GLM-4-Plus as requested

fix: qr detection reliability and log piping

feat: stabilize openclaw with rolling backups, blocking QR wait, and structured logging

Implement complete debug loop - atomic persistence, QR detection, structured logging

wa-guardian: skip 401 logout for 90s after 515 to avoid wiping just-saved creds

fix: use Space restart instead of config.apply after 515 to avoid double SIGUSR1

fix: guardian clears invalid session (401) so user gets fresh QR without relink

fix: do not remove WhatsApp credentials on restore; trigger config.apply on 515

fix: re-enable wa-login-guardian to handle 515 after QR scan

fix: prevent bad dataset overwrite and keep restored config

fix: preserve WhatsApp credentials in dataset when local empty (401/logout)

fix: remove WhatsApp creds on restore to avoid Bad MAC / decrypt failure

fix: set session.scope=global so chat reply sessionKey matches UI (fixes reply only working once)

fix: add allowedOrigins for HF Space; add scripts/openclaw.json.default for zhipu config

fix: use fallback config when ZHIPU_API_KEY unset to avoid MissingEnvVarError

fix: remove wa-login-guardian to fix 515 restart race condition