Multi Agent
15 articles about multi agent.
Why Passing Full Context Between Agents Fails
When you hand off full context between AI agents, the receiving agent latches onto whatever is emphasized and ignores the rest. Here is how to structure agent handoffs that actually work.
Error Propagation in Multi-Agent Networks - The Problem Nobody Talks About
When one AI agent makes a bad decision, every downstream agent inherits that error. Multi-agent systems amplify mistakes instead of catching them. Here is why error propagation is the real challenge.
Data Consistency Across Multiple Independent AI Agents
Running 5+ parallel AI agents on the same codebase creates file locking and conflict resolution challenges. Here is what works and what does not.
Running 5 AI Agents on the Same Codebase Without Branch Isolation
Lessons from running 5 Claude Code agents in parallel on a Swift, Rust, and Flutter desktop app. Same repo. Same branch. No isolation.
Building a Gateway Daemon for Claude Code Multi-Agent Scheduling
Using tmux sessions with individual agents plus launchd for scheduling. The hardest part of multi-agent orchestration is knowing when to intervene.
Git Worktrees Are the Secret to Running Multiple AI Agents Safely
Without isolation, parallel AI agents edit the same files and create merge conflicts. Git worktrees give each agent its own working directory on a separate branch.
Multi-Agent Hype vs Economic Reality in Production
A planner-executor-reviewer agent chain sounds elegant but burns 3x the tokens of a single well-prompted agent. Here is when multi-agent is worth it and when it is not.
Screenshots Are Better Than LLM Self-Reports for Multi-Agent Verification
Judge-reflection patterns in multi-agent systems sound good but the judge LLM can be fooled. Screenshots provide ground truth for verifying whether an action actually changed the screen.
The Consensus Illusion - When Multiple AI Agents Work on the Same Codebase
Five agents on the same branch with no isolation creates the illusion of a stable codebase. Why consensus fails and conflict resolution should be left to humans.
Anchoring Bias in Multi-Agent Systems - When One Agent's Output Biases All the Others
How anchoring bias silently degrades multi-agent AI systems when one agent's partial output influences the rest, and what you can do about it.
Parallel AI Agents Only Work with Genuinely Isolated Tasks
Running 5 AI agents in parallel sounds great until they step on each other's files. The key to parallel agents is genuinely isolated tasks with zero overlap.
Running 10+ Claude Code Agents Without Chaos - Tmux Orchestration
How to run ten or more Claude Code agents in parallel using tmux sessions for isolation, with practical tips on session management, context separation, and avoiding conflicts.
Start with One Agent, Not a Team - Why Single Agents Beat Multi-Agent Orchestration
A single well-scoped agent with real execution capability beats a complex multi-agent system. Multi-agent adds coordination overhead, error propagation, and debugging complexity you do not need yet.
What Running Parallel AI Agents Feels Like - Three Tiny Wars
Running multiple AI agents in parallel feels like managing three tiny wars at once. Each agent has its own front, its own problems, and its own momentum.
The Agent-to-Agent Economy Needs Agents That Can Actually Control a Computer
Everyone is talking about agent-to-agent communication. But the bottleneck is simpler - agents still cannot reliably control a single computer. Desktop control comes first.