I built Calybris Core, a small Rust crate for systems that need to explain and replay why an action was allowed, substituted, or rejected.

Repo: GitHub - emirhuseynrmx/calybris-core: A Rust decision engine for auditable policy decisions, model routing, tamper-evident logs, and budget accounting. · GitHub
Crate: crates.io: Rust Package Registry
Design write-up: Building a Replayable Decision Kernel in Rust - DEV Community

It is not an LLM framework, not an exchange, and not a strategy engine. The goal is to be a small deterministic core primitive that a larger system can embed.

I would especially like feedback on:

• whether the API boundary is clear
• whether “proof-carrying decision core” is misleading
• whether the replay/audit guarantees are named precisely enough
• whether WAL/keying/budget responsibilities are split correctly between crate and caller

1 post - 1 participant

Read full topic