FateLink/Coding-Ai

Update self_improving_ai_system/.env

Add synthetic data generator using API models as teachers

Update IMPROVEMENTS.md with new learning loop docs

Upload v2/layer5/training_orchestrator.py

Add composite reward function with format gating (DRIVE + ReflexiCoder style)

Add TAROT-style adaptive curriculum engine

Upload v2/layer4/core/adversarial_tester.py

Add Mistake Book experience replay buffer (Code-A1 style)

Upload v2/layer2/async_execution_engine.py

Add v0.2.0 improvements documentation

Add v0.2.0 defaults config with new features

Upload v2/layer2/llm_client_v2.py

Add fixed LLM client with unified backend interface - v0.2.0

Upload v2/layer4/core/ldb_debugger.py

Upload v2/layer4/core/grpo_trainer.py

Add critic model, reflexion memory, secure sandbox, persistent budget, unified LLM backends - v0.2.0

Upload v2/layer4/core/reflexion_memory.py

Upload v2/layer2/persistent_budget.py

Upload v2/layer2/secure_sandbox.py

Add unified LLM backend interface (Anthropic/DeepSeek/OpenAI) - v0.2.0

Upload 204 files

initial commit