🧠 MemPalace AGI Enhanced

A minimal self-improving agent that scores its own behavior and adapts over time.

---

⚡ What This Is

Most “AI agents” just generate outputs.

This one:

• takes actions
• evaluates itself
• adapts over time

A live feedback loop you can actually watch.

---

🔁 Core Loop

Action → Outcome → Self-Critique → Score → Adapt → Repeat

---

🧪 Example Output

[ITER 7] action=consolidate
[AGI] success: stabilized memory
[CRITIC] score=2 total=-4

📈 SCORE TREND:
1: -1
2: -2
...
7: -4

👉 The agent discovers beneficial actions.

---

🚀 Run It

python3 -m mempalace.learning_loop

---

🧠 Why It Matters

This shows a real feedback signal, not just output generation.

---

⚠️ Current Limitation

• Oscillates, not fully learning yet

---

🔮 Roadmap

• adaptive policy
• memory-driven decisions
• multi-agent dynamics

---

🧠 TL;DR

A tiny system that acts, judges itself, and evolves.