HN Showcase
Back to browse
GitHub Repository

A governed hybrid cognitive architecture in which neural intelligence is treated as a capability, not an authority.

2 starsPython

Elia – Governed hybrid architecture (LLM is capability, not authority)

by JMC-FR·Mar 7, 2026·2 points·2 comments
Visit ProjectView on HN

AI Analysis

●MidBig BrainBold Bet

LLM governance framework, but early-stage spec with no working code—Phase 0 skeleton promised.

Strengths
  • •Novel positioning: neural as capability, not authority—contrasts with mainstream agentic AI
  • •Thoughtful safety principles (graceful degradation, audit trails, circuit breakers)
  • •Seeks feedback from AI safety community, suggesting collaborative design intent
Weaknesses
  • •No implementation, no prototype, no working demo—only architectural specification and design docs
  • •Early-stage seek for help indicates project is pre-MVP; unclear if governance model actually scales
Category
AI/ML
Target Audience

AI safety researchers, systems architects, engineers building safety-critical AI systems

Similar To

Anthropic's Constitutional AI · MIRI's agent foundations research

Post Description

Hi everyone,

Elia: governed hybrid neuro-symbolic architecture where neural intelligence is a capability, never the authority.

Symbolic governance always in charge — LLMs are optional & validated.

Key features: strict separation of concerns, graceful degradation, audit trails, circuit breakers, heartbeats, SLOs, anti-deadlock.

Early stage (detailed spec in EL-ARCH.md). Phase 0 (minimal skeleton) coming soon. Not a full-time coder — open to implementation help.

Seeking feedback from systems architects, AI safety people, engineers: - governance model viability - biggest risks / blind spots - good first prototype domain (medical, monitoring, assistants, agents, education, ... any safety-critical domain)

Repo: https://github.com/Jmc-arch/elia-governed-hybrid-architectur...

Thanks for any thoughts! Jmc

Similar Projects