Fix Docker: pin python:3.10-bookworm with amd64 platform to fix registry manifest errors

Fix Docker build: use full python:3.10 image, add missing COPY for server/ and pyproject.toml

Major env overhaul: opponent negotiates naturally, gentler time penalty, relative aggression, simplified agent

Smart offer clamping + accept guard + softened time penalty + adaptive concession cap

Refined prompt: multi-round negotiation with valuation guardrail, temp 0.3

Optimize baseline prompt: multi-round negotiation strategy, all 3 tasks now pass

Cleanup: remove unused C++ files, test artifacts, and dev scripts

Add pyproject.toml, server/app.py, uv.lock — openenv validate now passes

Fix [END] format: restore score= field, add flush=True to all log prints

Fix inference.py stdout format: strict [START]/[STEP]/[END] compliance, all debug to stderr, try/finally for [END]

Add OpenEnv server (app.py) — expose /reset /step /state /ws /health endpoints

Add HF Spaces frontmatter with openenv tag + fix Dockerfile

docs: Refactor README to adhere to professional Hackathon conventions

feat: complete OpenEnv compliance, tasks, and logic fixes

feat: implement negotiation environment wrapper and LLM-based inference script

llms working

feat: implement Opponent class and add simulation testing framework for negotiation environment

Initialize project structure and core architecture interfaces

Initial commit