Feedback wanted: Kotlin/JVM runtime API and Evidence OS model

[morning-verlu]

I am looking for focused feedback from Kotlin/JVM developers, JVM backend maintainers, and people building agent infrastructure.

KAI OS is a local-first Evidence OS for AI agent runs in Kotlin. It is not trying to be another Kotlin LangChain clone. The runtime model is closer to infrastructure:

Agent    = Process
Workflow = Scheduler
Tool     = Syscall
Memory   = Process state
Run      = Evidence

The Kotlin/JVM evaluation path is here:

https://github.com/morning-verlu/KAI/blob/main/docs/KOTLIN_JVM_EVALUATION.md

Fastest no-install surface:

https://morning-verlu.github.io/KAI/evidence-viewer.html

Embeddable Kotlin runtime example:

./gradlew -p examples/kotlin-runtime-api run

What I would most like feedback on:

1. Does the runtime API feel idiomatic Kotlin, or does it feel like a translated framework from another ecosystem?
2. Is Agent = Process / Workflow = Scheduler / Tool = Syscall a useful model for JVM teams building agent infrastructure?
3. Are capability grants and syscall ledgers the right boundary for tools?
4. Would replayable capsules help PR review, CI, support, or audit workflows?
5. What would stop you from running kaios review against a real JVM repository?

Useful links:

• Kotlin/JVM evaluation path: https://github.com/morning-verlu/KAI/blob/main/docs/KOTLIN_JVM_EVALUATION.md
• Kotlin Runtime API: https://github.com/morning-verlu/KAI/blob/main/docs/KOTLIN_API.md
• JVM service review example: https://github.com/morning-verlu/KAI/tree/main/examples/jvm-service-review
• Comparison with Koog / LangChain4j / LangGraph / OpenAI Agents: https://github.com/morning-verlu/KAI/blob/main/docs/COMPARISON.md
• Contributor Board: https://github.com/morning-verlu/KAI/blob/main/docs/CONTRIBUTOR_BOARD.md

If you only have 5 minutes, opening the Evidence Viewer and saying what is confusing is already useful.