Launching today
Trigora
Durable execution without history replay.
1 follower
Durable execution without history replay.
1 follower
Trigora is a durable execution platform for long-running AI agents and programs. It lets applications survive crashes, waits, retries, and external events without replaying completed execution history. Built on the TCC engine, Trigora includes TypeScript, Python, and Rust SDKs, a local runtime and CLI, Trigora Cloud, plus built-in benchmarking and fault-injection verification.
I originally shared the TCC architecture before there was an engine or product. Since then, I built the portable TCC engine in Rust/WASM, the Trigora CLI and local runtime, TypeScript/Python/Rust SDKs, and Trigora Cloud.
The core idea is durable execution from committed state instead of rebuilding execution by replaying history, which keeps recovery extremely fast even as execution history grows.
I also wanted durability to be something developers can actually test, not just trust. `trigora bench` measures healthy-run performance, while `trigora verify` injects crashes after durable boundaries and checks that recovery matches the baseline without duplicate or missing effects.
Curious to hear how others building long-running agents or workflows are handling durability and recovery today.