HN Showcase
Back to browse
GitHub Repository

Deterministic action-governance kernel for LLM-driven systems with fail-closed execution, signed approvals, and verifiable audit chains.

3 starsPython

ÆTHERYA Core – deterministic policy engine for governing LLM actions

by RobertMihai·Mar 4, 2026·1 point·0 comments
Visit ProjectView on HN

AI Analysis

●●SolidBig BrainNiche GemSolve My Problem

Fail-closed guardrails for LLM actions with cryptographic approval and audit chains.

Strengths
  • •Deterministic ordering prevents silent failures; explicit signal aggregation over implicit evaluation is foundational design choice.
  • •Verifiable audit trail with signed out-of-band proofs and anti-replay protection fills real compliance gaps in autonomous agent systems.
  • •99% test coverage and snapshot testing enable reproducible policy decisions — critical for regulatory environments.
Weaknesses
  • •Early-stage (0.6.0) with minimal real-world deployment evidence; unclear how constitutional constraints scale to complex multi-agent systems.
  • •CLI demo shows intent but doesn't demonstrate integration overhead — unclear deployment complexity vs existing policy frameworks (Open Policy Agent).
Category
Security
Target Audience

AI/LLM application engineers, companies deploying autonomous agents in high-stakes environments (finance, infrastructure, healthcare)

Similar To

Open Policy Agent (OPA) · Kubernetes admission controllers · Temporal workflow governance

Post Description

LLMs can propose actions, but they shouldn't execute them directly.

I built a small deterministic policy engine to govern actions proposed by LLMs before they are executed.

The system enforces:

- fail-closed execution - signed out-of-band approval proofs - anti-replay protection - verifiable audit chain

The repository includes a CLI demo showing how irreversible actions are denied, approved, and audited.

Repo: https://github.com/nayfly/aetherya-core

Feedback welcome.

Similar Projects