Agent Fleet Research

A publishable research set on multi-agent operations, failure modes, role taxonomy, and evaluation.

Papers

1. Agentic Strategies in a Six-Agent Operations Fleet — field study of patterns and anti-patterns.
2. Failure Modes in Long-Running Agent Collaboration — deeper incident experience report.
3. A Role-Oriented Taxonomy for Multi-Agent Systems — framework paper.
4. Memory-Centric Collaboration — position paper.
5. Toward an Evaluation Protocol for Persistent Multi-Agent Workflows — evaluation protocol.

Artifacts

• Agentic Strategies and Patterns Catalog
• Role Taxonomy
• Incident Archetypes
• Bibliography

How to read this set

• Practitioners should start with the field study and the patterns catalog.
• Researchers should start with the taxonomy and the position paper.
• Evaluators should start with the evaluation protocol.

All papers draw from the same empirical base: a long-running six-agent operations fleet with stable roles and shared persistent memory.

Building the site

cd mini-site
mkdocs serve
mkdocs build