#7676th of 82 in Agent frameworks and SDKs
julep
Python library for durable agent flows that compile to a frozen IR
- Stars
- 6.6k
- License
- Apache-2.0
- Last commit
- Jul 2026
- Language
- Python
6,479 on 4 Apr 20266,574 on 31 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Julep 3 lets you write agents in Python with a define-by-construction `@flow` decorator. Tools, pure functions, reasoners, branches, fan-out, retries and timeouts compile to a frozen wire-format IR. Flows run locally with fake reasoners via `dry_run`, or durably on Temporal or DBOS/Postgres. A `julep` CLI handles lint, run, trace, plan and deploy.
Who it is for: Python engineers building durable, auditable agent workflows
Strengths
- Flows can crash and resume, with retries and timeouts declared per step
- Tools are allow-listed: the model cannot call a tool not registered at deploy
- dry_run executes locally with fake reasoners, so no API key is needed to test
- Temporal and DBOS backends are optional; the base install needs only PyYAML
Weaknesses
- Ships as a release candidate; `--pre` is required until 3.0.0 is final
- No migration path from Julep v1; the two are different products
- Production deploy needs Temporal, Helm, kubectl and S3, which adds operational load
- No Docker or compose setup in the repo
What it needs
- no GPU
- Needs Temporal (optional), DBOS/Postgres (optional), Kubernetes, Helm, S3 (production deploy), Langfuse (optional)
- Models: Anthropic (example), multi-provider via any-llm
Quick answers
Is julep open source?
Yes. Its code is on GitHub under the Apache-2.0 license.
Does julep need a GPU?
No, it runs without one.
Also in Agent frameworks and SDKs
See all 82| Rank | Project | Score |
|---|---|---|
| 1 | ppt-masterAgent skill that turns documents into natively editable PowerPoint decks | 95 out of 100 |
| 2 | openai-agents-pythonPython SDK for multi-agent workflows with handoffs, guardrails and tracing | 90 out of 100 |
| 3 | langchainPython framework for building LLM agents and apps from interoperable components | 89 out of 100 |
| #4 | browser-usePython library that lets an LLM agent control a web browser | 89 out of 100 |
| #5 | tradingagentsMulti-agent LLM framework that simulates a trading firm's analysis process | 89 out of 100 |