julep-ai/julep

Project overview

Julep takes a structurally different approach to agent engineering by compiling workflows into a durable wire-format intermediate representation, enabling crash recovery and step-level explainability that ad-hoc loop patterns cannot provide.

Project type

AI Agent · Workflow

Use cases

Automation

Deployment

Refer to project documentation

License

Apache-2.0

Best for

• Developers and AI engineers who need durable, explainable, retry-safe agent workflows.

Key capabilities

• Agents are constructed as composable, durable dataflows that can crash, resume, retry safely, and explain every step.

Limitations and risks

• Julep currently ships as a release candidate rather than a stable general-availability release.
• Julep 3 is a ground-up rewrite with no migration path from v1; the two versions are different products.

Getting started

• Install the release candidate with pip install --pre julep and follow the documented ten-minute quickstart.

Evidence and sources

• GitHub project description: Julep — durable, composable AI agents. Flows that crash and resume, retry safely, and explain every step.
• README: Julep builds agents as **composable, durable dataflows** instead of ad-hoc loops: flows can crash and resume, retry safely, explain every step through a derived projection, and de…
• README: Julep 3 currently ships as a release candidate, so the `--pre` flag is required; it drops once `3.0.0` is final.
• README: 10 minutes, no API key. Install the base package and run this as a normal Python script:
• README: `julep` is the developer CLI for a whole **module of agents** — "dbt for agents, terminal-native."

Reviewed at: 2026-07-16T13:11:08Z