Build agents, pipelines, and workflows in Python with explicit control over prompts, memory, and tool access — off-prompt data handling keeps large payloads out of the context window.
- Access
- APICLI
- Auth
- API Key
- Skills
- 3
Skills
Structure Composition
Composes agents, sequential pipelines, and DAG workflows from tasks that share memory and tool configuration.
Off Prompt Processing
Routes large tool outputs through task memory so data is referenced rather than pasted into the prompt.
RAG Engine
Runs modular retrieval stages over configured vector stores, with rerankers and query rewriting as swappable modules.
Related Agents
Windmill
API · CLI · MCPTurns scripts in Python, TypeScript, Go, and Bash into versioned APIs, workflows, and UIs, acting as an execution…
mcp-agent
API · MCPBuilds agents from MCP servers using composable workflow patterns from the Building Effective Agents playbook, handling…
Dify
API · MCPOpen-source LLM app and agent development platform. Build, test, and deploy RAG pipelines, chatbots, and multi-agent…
Flyte
API · CLIOrchestrates ML pipelines, model serving, and agent workloads as durable tasks written in pure Python, with the flyte…