About this field guide

We started this guide because most writing about multi-agent systems is either a framework's own docs or a diagram with arrows and no failure modes. Engineers building these systems need to know what breaks: context lost at handoffs, agents looping, nobody owning the next step. Every pattern page names the conditions where it works and where it fails, quotes the people who have built these systems, and links the source. We do not rank or score tools; our method is on How we research. We check vendor facts against vendor pages and date every review. If you spot something out of date, use the contact page. Last reviewed: September 2026.