agnosticarchitect/ace-os-v8 — explained in plain English
Analysis updated 2026-05-18
Run a role-playing simulation where character actions must pass strict rule checks
Build a custom character using an Entity_DNA file with defined stats and traits
Track a simulation's changing state in a Current State Registry file across a session
Study the included Bleach anime demo to learn the framework before building your own
| agnosticarchitect/ace-os-v8 | 16nic/comfyui-agnes-ai | 521xueweihan/hgdoll | |
|---|---|---|---|
| Stars | 19 | 19 | 19 |
| Language | — | Python | Kotlin |
| Setup difficulty | moderate | moderate | hard |
| Complexity | 3/5 | 2/5 | 4/5 |
| Audience | general | vibe coder | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires a cloud API model or a local model of at least 32 billion parameters, smaller models are not supported.
ACE OS v8.0 (Agnostic Causality Engine) is a structured framework of text files and templates designed to make large language models run role-playing simulations in a strict, rule-bound way. Instead of letting an AI improvise freely, this system forces it to behave like a sequential logic processor: every action either passes or fails based on defined rules, and there is no room for narrative shortcuts or "plot armor." The core idea is that everything in a simulation, whether a character, a location, or a rule set, is stored as a separate, swappable module the framework calls a data cartridge. A character is described in an Entity_DNA file that lists physical constants and power levels as numbers. A Memory Engrams file stores psychological anchors and behavioral triggers. A Field Rules file sets environmental limits for a specific setting. A Current State Registry JSON file tracks everything changing in real time during a session. Because these components are isolated from each other, you can swap a character mid-session or move from one fictional world to another without losing the history of what happened. Three hard constraints govern every output. ZERO_BRIDGING means the model cannot produce conversational intros or meta-commentary, it outputs actions and dialogue directly. COT_ENCAPSULATION means all the internal processing and comparisons stay hidden in the model's reasoning layer and never appear in the output. ABSOLUTE_CAUSALITY means if a character's action potential is below the environmental resistance, the action is blocked with a hard verdict, not narrated around. The repository includes a full demonstration using a character from the Bleach anime universe, complete with pre-filled Entity_DNA, memory, field rules, and two save-state snapshots taken 30 turns apart so you can see the system in operation before building your own. The framework is rated for cloud API models and local models at 32 billion parameters or above. Smaller models are listed as not compatible. The project is a static artifact under the MIT license, with no support, no issue tracking, and no plans for updates. Users who want changes are expected to fork it independently.
A prompt-engineering framework of text templates that forces large language models to run rule-bound role-playing simulations with hard pass or fail action checks instead of free improvisation.
Released under the MIT license, free to use, modify, and redistribute, including for commercial projects.
Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.
Mainly general.
This repo across BitVibe Labs
Verify against the repo before relying on details.