docs: update README with badges, personaxis explanation, MIT license

- Add GitHub, HuggingFace Space, and Hackathon badges
- Explain personaxis.md vs PERSONA.md (spec vs live system prompt)
- Add 10-layer list and SLM persona concept from launch thread
- Change license to MIT in metadata and license section
- Add assets/ folder with demo video

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

README.md

@@ -5,7 +5,7 @@ colorTo: indigo
 sdk: docker
 app_port: 7860
 pinned: true
-license: apache-2.0
+license: mit
 short_description: Governed self-evolving AI persona, fully local on a SLM.
 tags:
   - build-small-hackathon
@@ -20,21 +20,46 @@ tags:
 
 # Daimon
 
+[![GitHub](https://img.shields.io/badge/GitHub-davidquicast%2Fdaimon-black?logo=github)](https://github.com/davidquicast/daimon)  [![🤗 Space](https://img.shields.io/badge/🤗%20Space-davidquicast%2Fdaimon-FFD21E)](https://huggingface.co/spaces/davidquicast/daimon)  [![🤗 Hackathon](https://img.shields.io/badge/🤗%20Build%20Small-Hackathon-blue)](https://huggingface.co/spaces/build-small-hackathon/daimon)
+
 **A governed, self-evolving AI persona that runs fully local on a small model.**
 
-For this hackathon, Daimon hands its own `personaxis.md` to **MiniCPM5-1B, running fully
-local through llama.cpp** - under 1B parameters, no cloud. Every message, MiniCPM appraises
-the conversation and proposes small changes to its own personality and mood, and a
-governance gate clamps, audits, or rejects each one live. You watch the 10-layer vector
-move, watch `PERSONA.md` rewrite itself, and watch it refuse to touch its own identity.
+> Can SLMs become a dedicated persona layer for AI apps? Rather than static prompts, Daimon explores whether persona state, evolution, and constraints can operate independently, while other models focus on reasoning and tasks.
+
+Daimon is an experimental implementation of the [personaxis](https://github.com/personaxis/persona.md) 10-layer persona architecture, running on **MiniCPM5-1B-Q4_K_M via llama.cpp**, under 1B parameters, no cloud.
+
+Every message, MiniCPM appraises the conversation and proposes small changes to its personality and mood. A governance gate clamps, audits, or rejects each change live. You watch the vector move, `PERSONA.md` rewrite itself, and the model refuse to touch its own identity.
 
 ```
 chat -> appraisal (JSON constrained by grammar) -> deterministic mapping
      -> governance + clamp -> recompile PERSONA.md -> memory
 ```
 
-Track: **Thousand Token Wood**. Lanes: **OpenBMB (MiniCPM)** + **OpenAI Codex**.
-Badges: **Tiny Titan** (<= 4B), **Off-the-Grid** (all local), **Off-Brand** (custom UI), **Best Agent**.
+### How the persona works
+
+`personaxis.md` is the **architecture spec**: a structured definition of the persona across 10 governed layers. It declares who Daimon is, what it values, how it reasons, and what it cannot change about itself.
+
+`PERSONA.md` is the **live system prompt**: compiled from that spec after every interaction. It is what the model actually reads. When the governance engine accepts a change, it rewrites `PERSONA.md` so the next message already talks to a slightly different Daimon.
+
+**The 10 personaxis layers:**
+1. Identity
+2. Character
+3. Personality
+4. Values & Drives
+5. Affect
+6. Cognition
+7. Memory
+8. Metacognition
+9. Reflexive Self-Regulation
+10. Persona
+
+---
+
+## Demo
+
+<video src="assets/demo.mp4" controls width="100%"></video>
+
+> Posted on X: [x.com/davidquicast/status/2066722575577706745](https://x.com/davidquicast/status/2066722575577706745)
 
 ---
 
@@ -42,9 +67,9 @@ Badges: **Tiny Titan** (<= 4B), **Off-the-Grid** (all local), **Off-Brand** (cus
 
 ```
 .personaxis/     spec: project baseline + live "daimon" persona
-engine/          living loop (appraise -> map -> govern -> recompile -> memory)
+engine/          living loop (appraise → map → govern → recompile → memory)
 model/           small-model serving (llama.cpp + MiniCPM GGUF)
-app/             single gr.Blocks UI at "/" (Off-Brand)
+app/             single gr.Blocks UI at "/"
 AGENTS.md        project guidance
 PERSONA.md       compiled behavioral baseline
 ```
@@ -68,4 +93,6 @@ Open `http://localhost:7860/`. No cloud APIs required.
 
 ## License
 
-The `persona.md` spec and these personas are public. See the sibling `persona.md` repo for the spec itself.
+MIT. The specification belongs to the community. Personaxis builds the tooling and platform around it.
+
+See [personaxis/persona.md](https://github.com/personaxis/persona.md) for the spec.

assets/demo.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0080492c7b53c8c4b2ff3c58fd5bc188a11ee143a6f095140efaa3f931bab186
+size 33045122