Alexis Diamond Empire Prototype

Deterministic, auditable prototype of the "Autonomous Invariant Intelligence" loop. It targets decidable domains (propositional logic) to guarantee zero-entropy outputs via a Z3 oracle.

---

Constitutional Framework

This project operates under the Constitution of a Deterministic Assistant (CDA-v1.0), establishing transparent operational boundaries for AI systems.

Core Principles:

• 🔍 Transparency: Glass Box Mandate - All operations are auditable
• ⚙️ Determinism: Predictable Tool Mandate - No autonomous agency
• 🤝 Subservience: Tool-in-Hand Mandate - Human authority paramount

See .github/CONSTITUTION.md for full constitutional text and .github/COMPLIANCE.md for implementation details.

---

Features

• Deterministic proposal loop (no sampling)
• Hamiltonian-style barrier on potential energy (V = 1 - coherence)
• Z3-backed coherence oracle for tautology checking
• Immutable JSONL audit ledger of every state transition

Quickstart

pip install -r requirements.txt
python AlexisDiamondEmpire.py --symbols p q

Outputs the converged candidate and writes an audit log to audit-ledger.jsonl.

How it works

1. Propose a candidate formula deterministically from a fixed list.
2. Evaluate coherence via the oracle (1.0 if tautology; 0.0 otherwise).
3. Compute potential energy V = 1 - coherence.
4. Hamiltonian filter accepts if V <= epsilon; otherwise transmute to the next candidate.
5. Every step is written to the append-only ledger for reproducibility.

Extending

• Add more candidates in AlexisDiamondEmpire._default_candidates.
• Adjust epsilon or max_steps via CLI flags.
• Swap the oracle to another decidable domain by implementing InvariantOracle.coherence.

Dependencies

• Python 3.11+
• z3-solver (see requirements.txt)