AbsoluteAGI

Still early: runtime implementation is evolving. This release focuses on architecture & documentation to invite discussion. Star the repo if you're interested in building persistent artificial agents! github.com/AbsoluteAGI/Ho…

Other highlights: - Motivational drives (desire/emotion/preference/confidence/fatigue) influence attention & decisions - Body channels unify interaction - Sleep cycles consolidate memory & compile cognitive patterns - Recursive PEOR ensures robust long-horizon task execution

Memory system: - L1: Working memory (short-horizon active tasks) - L2: Narrative memory (first-person continuity) - L3: Knowledge memory (world, social, preferences, causality) - L4: Skill memory (successful plan templates) - L5: Identity / Personality parameters

Cognitive loop highlights: - Dual paths: external stimulus & internal ideation - Attention-centric execution - Commitments handled via recursive PEOR loops (Plan→Execute→Observe→Replan) - Step outcomes recorded into L2/L3 and distilled into L4 skills

Key architecture overview: L0 – Existence Model L1 – Data Contracts L2 – Runtime Processes L3 – Cognitive Pipeline L4 – Memory Architecture L5 – Body & Perception L6 – Desire, Emotion, Preference L7 – Sleep & Compilation L8 – Initialization & Birth

We just open-sourced HomoErectus: a cognitive operating system for persistent artificial agents. It goes beyond traditional AI, supporting first-person narrative continuity, autonomous ideation, and adaptive learning. Docs & architecture: github.com/AbsoluteAGI/Ho…