AgentForge

Why we included this project

AgentForge is a low-code Python framework for building autonomous agents: you describe agents, personas, and multi-step workflows in YAML files instead of writing orchestration glue by hand. Its declarative Cogs handle branching logic and shared memory across agents, so a project can start with a single smoke-test agent and grow into a coordinated multi-agent pipeline without a rewrite. You can also point different agents at different models, mixing OpenAI, Gemini, Claude, or local models through Ollama and LMStudio, which suits teams that run some workloads on-premise. The built-in memory and prompt editing are worth a look if you need long-running, context-aware agents you can tune without restarting.