julep
Julep — durable, composable AI agents. Flows that crash and resume, retry safely, and explain every step.
Julep is a Python framework for building durable, composable AI agents as dataflows that can crash and resume, retry safely, and explain every execution step. Unlike ad-hoc agent loops, Julep compiles flows from ordinary Python into a frozen intermediate representation with built-in error handling, timeouts, retries, and MCP tool integration. It supports both local dry-run execution and production deployment on Temporal or DBOS, with optional self-hosted control plane for secrets management and multi-agent orchestration. Developers author agents using the @flow decorator, register tools and reasoners, and deploy with the julep CLI for testing, tracing, and production releases.
Key Features
Use Cases
- 01Building customer support agents with ticket lookup, knowledge base integration, and structured reply generation
- 02Creating multi-step research workflows that can resume after API failures or rate limits
- 03Orchestrating AI pipelines with MCP tools for memory systems, databases, and external APIs
- 04Deploying production agent systems with encrypted secrets and compliance-ready audit trails
- 05Testing agent behavior locally with deterministic mocks before live deployment
- 06Managing multi-agent applications with cross-agent dependencies and shared tool surfaces
Related Agents
View morehermes-agent
The agent that grows with you
agency-agents
A complete AI agency at your fingertips - From frontend wizards to Reddit community ninjas, from whimsy injectors to reality checkers. Each agent is a specialized expert with personality, processes, and proven deliverables.
openinterpreter
A coding agent for open models like Kimi K3
cline
Autonomous coding agent as an SDK, IDE extension, or CLI assistant.
julep — FAQ
What is the Julep AI sub-agent framework?+
Julep is a Python framework for building durable, composable AI agents as dataflows rather than loops. Agents can crash and resume, retry failed steps safely, and integrate MCP tools with frozen schema validation.
How do I install Julep?+
Run 'pip install --pre julep' to install the base authoring and compile package. Add extras like 'julep[temporal]' for durable execution, 'julep[mcp]' for MCP integration, or 'julep[server]' for the self-hosted control plane.
Does Julep require API keys or external services?+
The base package requires no API keys and supports local dry-run execution. Production deployment optionally uses Temporal or DBOS for durable execution, and LLM reasoners require provider API keys (Anthropic, OpenAI, etc.).
Which AI clients work with Julep agents?+
Julep is a standalone agent framework, not a plugin for existing AI clients. It builds independent agents that integrate with MCP servers and can be deployed via its own CLI and control plane.
Is Julep free and open source?+
Yes, Julep is open source under the Apache-2.0 license. Version 3 is currently in release candidate status and available for use.
How do I deploy a Julep agent to production?+
Use the julep CLI to test locally with 'julep run', then deploy with 'julep deploy <agent> --env <environment>'. Production requires configuration in pyproject.toml for Temporal/DBOS connection, worker images, and secrets management.
How do I install julep?+
Open the source repository on GitHub and follow its README. julep is a agent — MCP Agents Market links you directly to the official repo.
Is julep free?+
julep is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.