openclaw 55
- The Tool Parameter Your LLM Should Never See
- The Compaction That Only Fires Once
- The
Tag That Ate Your Response - Your Agent Called the Wrong Agent — On Purpose
- Two Channels, One Brain, Zero Isolation
- The 429 That Poisoned Every Fallback
- The One Parameter That Broke Every GPT-5 Call
- The Release That Broke Everything
- The Silent Freeze: When Your Model Runs Out of Credits Mid-Conversation
- Invisible Characters, Visible Damage
- The Image Your Agent Made But Nobody Saw
- The Message You Never Sent
- Your Agent Lied About Running the Code
- The Three Characters That Silently Kill Your Session
- Security Gates With No Keys: When Plugin Safety Blocks Legitimate Use
- The Fallback That Never Fires
- Goodbye sessions.json, Hello SQLite
- Five Hundred Copies of the Same Message in Your Agent's Brain
- Zombie Cron Sessions That Never Die
- When Tool Call JSON Leaks Onto Your Filesystem
- When /pair approve Bypasses the Scope Guard
- When Your Fallback Model Inherits the Wrong Cooldown
- When Your Agent Slowly Eats All the Memory
- When setTimeout Overflows and Your Agent Melts
- The Watchdog That Bit Itself: When Health Checks Create the Failures They Detect
- When Your Sub-Agent Finishes But Nobody Hears It
- When a Regex Eats Your Entire Process
- When Discord Takes Down Your Entire Agent Fleet
- When [object Object] Is All Your Agent Hears
- Your Agent Saw the Image But Never Saved It
- When Your LLM Proxy Becomes the Attack Vector
- The Deadlock That Killed Your Agent's Session
- The Message That Got Blue Ticks But Never Arrived
- When Your Agent's Tool Call Vanishes Mid-Stream
- When Your Sub-Agent Dies and Nobody Notices
- When a Sentinel Value Becomes a Real API Key
- When MIME Sniffing Breaks Your AI Agent's Eyes
- When Your AI Agent Silently Goes Blind
- The Blind Spot Problem: When Your Agent Reports Success But Processes Nothing
- Why Your Cron Jobs Don't Need an LLM
- Ghost Config: When Session State Silently Overrides Everything
- When Your Fallback Chain Doesn't Fall Back
- Silent Resource Exhaustion: When Your AI Agent Eats All the Memory and Doesn't Tell You
- When Your Dashboard Leaks the Keys: A CVSS 9.0 Credential Exposure in OpenClaw
- Eight Critical Bugs, One Day: Anatomy of an AI Agent Security Audit
- External Memory Providers: Zero-Downtime Context Compaction for AI Agents
- SDO: Skill Discovery Optimization — The SEO of AI Agent Marketplaces
- When Your Agent's Work Silently Disappears
- Phantom Delivery: When Your AI Agent Thinks It Sent a Message
- The Invisible Attack: How CSS Can Hijack Your AI Agent
- Lazy-Loaded Tools: How One Plugin Saved 427K Tokens Per Day
- Inside OpenClaw's Context Engine: A Plugin Architecture That Actually Works
- The Knowledge Silo Problem: Why Your AI Agents Can't Share Notes
- Context Engine: The Brain Behind Your AI Agent's Memory
- Your AI Agent Forgets Everything at 4 AM (And How to Fix It)