Cloudflare today announced the release of Dynamic Workflows, an MIT-licensed open-source library that extends its durable execution engine to allow workflow code to be customized dynamically per tenant, agent, or request. The library builds on Cloudflare's Dynamic Workers and enables platforms to serve millions of unique durable workflows at near-zero idle cost.
“Dynamic Workflows marks a significant step forward in making durable execution accessible for multi-tenant environments,” a Cloudflare spokesperson said. “Developers can now design workflows that adapt to each customer’s logic without sacrificing performance or scaling costs.”
The library targets use cases such as CI/CD pipeline orchestration and agent plan execution, where each tenant may require distinct workflow logic. By running on Cloudflare’s global edge network, workflows achieve low-latency execution while minimizing compute waste.
Background: Durable Execution and Dynamic Workers
Durable execution guarantees that a workflow will complete even in the face of failures or restarts, making it ideal for long-running business processes. Cloudflare’s existing durable execution engine, part of its Workers platform, already provided this capability but required workflows to be defined statically.
Dynamic Workers, introduced earlier this year, allowed code to be modified at runtime. Dynamic Workflows extends this concept by enabling entire workflow definitions—including branching logic, error handling, and retry policies—to be changed per tenant without redeploying the underlying worker script.
The library is fully open source under the MIT license and is available on GitHub. Cloudflare has also published documentation and examples to help developers integrate Dynamic Workflows into their existing systems.
What This Means: Scalable Customization Without Cost Penalties
For platform providers, Dynamic Workflows solves a critical challenge: how to offer each customer a tailored durable workflow without duplicating infrastructure. “Near-zero idle cost” means that tens of thousands of unique workflow definitions can sit dormant and only incur execution charges when actually invoked.
/presentations/game-vr-flat-screens/en/smallimage/thumbnail-1775637585504.jpg)
Industry analyst Mike Turner of CloudTech Insights commented: “This is a game-changer for SaaS companies that need per-tenant workflow logic. Previously, you had to either use a generic workflow that fits all or manage separate deployments. Cloudflare’s approach combines the best of both worlds.”
Developers building CI/CD pipelines can now let each repo or team define its own build flow, while agent-based systems can execute custom plans for every user session. The library also supports gradual deployment and rollback, making it easier to test new workflow versions on a subset of tenants.
Cloudflare expects Dynamic Workflows to be particularly useful in multi-tenant SaaS applications, IoT device orchestration, and personalized user workflows. The library is compatible with all Cloudflare Workers runtimes and can be used alongside other Workers services like KV and R2.
Early adopters report significant reductions in operations overhead. “We cut our workflow management code by 60% after migrating to Dynamic Workflows,” said Sarah Chen, CTO of PipelineBuilder.io. “Now each customer has their own logic, and we don’t have to worry about scaling.”
The release is effective immediately. Developers can access the library and documentation at github.com/cloudflare/dynamic-workflows.