experimental-research

Here are 3 public repositories matching this topic...

kar-ganap / ate-features

Agent Teams Eval: comparing Claude Code Agent Teams vs single-agent for feature implementation on LangGraph. Ceiling effect, 3.6x speedup, zero peer communication.

multi-agent-systems feature-implementation langgraph experimental-research claude-code agent-teams subagents agent-teams-eval

Updated Feb 26, 2026
Python

kar-ganap / ate-arch

Star

Agent Teams Eval: comparing Claude Code Agent Teams vs subagents for architecture design. First significant result — Agent Teams advantage d=+0.99, p=0.014.

software-architecture multi-agent-systems llm-as-judge experimental-research claude-code agent-teams subagents agent-teams-eval stakeholder-simulation