deepansh-saxena/DuraLang

Make stochastic AI systems durable with one decorator

35
/ 100
Emerging

This helps developers build AI agents that are resilient to failures like network timeouts or system crashes. It takes existing LangChain agent code and makes every individual LLM or tool call automatically recoverable and retryable. This ensures that long-running AI agent tasks, which might involve many steps, don't lose all progress if an intermediate step fails. It's designed for Python developers creating complex, multi-step AI agents using LangChain.

Available on PyPI.

Use this if you are developing LangChain-based AI agents and need them to reliably complete tasks without failing due to transient errors or system outages, automatically resuming from the last successful step.

Not ideal if your AI agent workflows are simple, deterministic pipelines that don't involve unpredictable LLM-driven decision making or long-running operations prone to external failures.

AI-agent-development reliable-AI-systems LangChain-workflow production-AI fault-tolerant-applications
Maintenance 13 / 25
Adoption 4 / 25
Maturity 18 / 25
Community 0 / 25

How are scores calculated?

Stars

8

Forks

Language

Python

License

MIT

Category

multi-agent

Last pushed

Apr 01, 2026

Commits (30d)

0

Dependencies

5

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/agents/deepansh-saxena/DuraLang"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.