Task execution
Submit work with priorities to a managed pool that runs it concurrently. Priority bounds keep urgent work first, and aging makes sure lower-priority work still gets its turn - so nothing waits forever, even when demand spikes.
A brainless slime mold rebuilt the Tokyo rail map in ~26 hours
Thryve: where your projects thrive.
A high-performance, open-source task-execution framework written in Rust and usable from virtually any language through its C-compatible FFI. It provides prioritized concurrency, an event-driven observer system, and a plugin architecture that extends the engine at runtime. It's free under the AGPL-3.0, and available under a commercial license for teams whose products can't be open source.
Three pillars, one engine
At the center is the Engine - takes in work, routes notifications, gathers results, and coordinates pause, resume, and shutdown. Three pillars sit on top of it.
Observer system
Components react to what happens without being wired directly to one another. Emit an event once and every interested part responds - fewer dependencies to maintain, and far less that breaks when the system changes.
Plugin architecture
Add or replace capabilities at runtime through plugins, with no changes to the core. Teams extend Thryve to fit their own systems and keep their modules cleanly separate from the engine.
Built for production
Practical building blocks that make the engine easy to integrate, operate, and observe. The engine itself is written in native Rust throughout.
FFI layer
A C-compatible interface exposes the engine to virtually any language - opaque handles, explicit create/free pairs, and clear ownership rules. Adopt Thryve without rewriting your stack in Rust.
TOML configuration
Workers, priority bounds, aging, observers, plugins, logging, and run mode are all declared in one readable file. Tune behavior in production without touching code.
Graph visualization
Task topology is captured and rendered as a graph, so teams can see how work actually flows through the engine and diagnose issues quickly.
Interactive & service modes
Run an interactive loop while developing, or run Thryve as a managed background service - including Windows service control, with start, stop, pause, and resume - in production.
Open at the core, licensed for the rest
Thryve is one engine under two licenses. Most teams use the open edition freely; those whose products can't meet its terms take a commercial license instead.
Open source
GNU AGPL-3.0
Free to use, study, modify, and share. If you distribute Thryve or offer it to users over a network, the AGPL asks that you release your corresponding source under the same license. Ideal for open-source projects, research, and internal evaluation.
Read the AGPL-3.0 →Commercial license
For proprietary & closed-source use
Ship Thryve inside a product you don't open-source, or run it in a commercial service, without the AGPL's source-disclosure requirements. Terms are scoped to how you deploy it. Send a proposal and we'll take it from there.
Not sure which one applies to you? A short proposal gets you a clear answer - no obligation.
Why the name
Teams that build on it thrive - that's the idea. It's also a quiet nod to Physarum polycephalum, the brainless slime mold that famously grew a near-optimal copy of the Tokyo rail network from nothing but oat flakes. Thryve borrows the same instinct: many workers do the work at the core while plugins and observers contain the data that needs processing.
Ready to run Thryve commercially?
Tell us how you're deploying it and what you need.