Agent DailyAgent Daily
releaseintermediate

[Release] openclaw/openclaw v2026.6.33: openclaw 2026.6.33

By RomneyDagithub
View original on github

OpenClaw v2026.6.33 is a major security and reliability release featuring safer network boundaries with response size caps and credential protection, more reliable long-running agents with improved stall detection, stronger channel delivery for Discord and Telegram, safer credential recovery mechanisms, and support for extended-stable package updates. The release includes 169 merged PRs addressing authorization, tool authority, external tooling scoping, and numerous stability fixes across gateway, provider, browser, and webhook systems.

Key Points

  • Implement response size caps for provider streams, Discord REST, browser fetches, and OAuth paths to prevent hostile response exploitation
  • Distinguish genuine agent stalls from active long model calls using improved liveness checks and watchdog semantics for reliable long-running operations
  • Fix Discord reconnection logic to prevent silent message drops and non-idempotent send repetition; improve Telegram bot-to-bot thread handling
  • Preserve SecretRef-backed credentials during service restarts and prevent OAuth repair from overwriting valid destination profiles
  • Enable extended-stable channel selection for package installations with proper availability notices instead of silent fallback behavior
  • Require explicit human/plugin approval for Codex app-server commands and bind exec auto-review to exact resolved commands
  • Use short-lived session-bound attach grants for external MCP loopback clients instead of inheriting mutable child-process authority
  • Redact secrets from MCP status output, reject disallowed browser origins before unauthenticated handling, and validate Gateway action requester provenance
  • Bound agent-run caches, fix file descriptor leaks in lock probes, preserve valid UTF-8 in close reasons, and handle transient filesystem races in heartbeat reads
  • Prevent Telegram token leakage across log chunks, reject replayed Twilio requests under sustained traffic, and tombstone corrupt queued channel rows

Found this useful? Add it to a playbook for a step-by-step implementation guide.

Workflow Diagram

Start Process
Step A
Step B
Step C
Complete
Quality

Concepts