Project ideas from Hacker News discussions.

Brood War Bench

📝 Discussion Summary (Click to expand)

1. StarCraft as a benchmark for LLM strategy and tactics
- “This is a great idea for a benchmark. Something all the benchmarks seem to be missing is strategy, tactical solutions …” – malfist
- “I predict LLMs will reach superhuman level and beat even that model in the next 12 months.” – benswerd
- References to GoBench and similar efforts show interest in measuring LLMs’ strategic ability via games. – GodelNumbering

2. Nostalgia and personal ties to the StarCraft community
- “I love StarCraft. I started playing it right from the beginning… I literally met people that have spread to almost every continent …” – pelagicAustral
- “Later we even listened to Eminem and played violent video games. Most of have never even been charged with a crime…” – sidewndr46
- “Oh boy, I've spent more hours playing it than I dare to admit. I won over 10 000 battle.net games …” – snicky

3. Technical challenges and limits for LLM agents in real‑time StarCraft
- “Starcraft is APM‑dependent. Unless the latency will improve greatly in frontier reasoning LLMs (which is unlikely), it will remain a bit like knitting with an excavator.” – orbital‑decay
- “Yeah but Starcraft needs, like, 10‑20x the APM these agents are doing.” – loeg
- “Playing these games autoregressively isn't even the right way to use the LLM for this task … it ends up on how much go transcripts ended up in the training…” – nullc


🚀 Project Ideas

Generating project ideas…

BWReplayLib: Open‑Source StarCraft Brood War Replay Parser & API

Summary

  • Provides a lightweight parser (Python/Rust) that converts .rep replay files into structured event streams (unit actions, build orders, APM, high‑level strategy tags).
  • Enables developers to quickly extract gameplay data for training agents or building analytics without reinventing the wheel.

Details

Key Value
Target Audience AI researchers, bot developers, replay analysts
Core Feature Parse .rep files into JSON/Protobuf with actions, timestamps, and optional strategy classification
Tech Stack Rust core with Python bindings (PyO3), optional FastAPI wrapper; Docker for deployment
Difficulty Medium
Monetization Hobby

Notes

  • HN commenters lamented the lack of a library to decipher BW replays (stackghost) and wished for easy observation APIs.
  • Provides a reusable building block that can accelerate agent harnesses and replay‑based benchmarks.

StarCraft LLM Agent Harness with Code Mode & Learning Loop

Summary

  • A ready‑made harness that connects LLMs to StarCraft Brood War via BWAPI, exposing screen‑state, unit info, and action commands as function calls.
  • Includes a “code mode” macro layer and a replay‑based learning loop so agents can improve from past games without manual retraining.

Details

Key Value
Target Audience LLM experimenters, AI hobbyists, competitive bot builders
Core Feature Observation/action API + pluggable LLM adapter (supports OpenAI, Claude, local models) + code‑mode macro executor + automated self‑play logging
Tech Stack Python (asyncio), BWAPI‑C++ wrapper via ctypes or pybind11, FastAPI for web UI, SQLite for replay storage
Difficulty High
Monetization Hobby

Notes

  • Commenters wanted a harness that lets agents learn from past games and integrate code mode (benswerd, shard972).
  • Addresses the frustration of building custom harnesses from scratch and gives a clear path to iterative improvement.

SC2Arena: LLM vs LLM Tournament Platform for Brood War

Summary

  • An online arena where users can upload their LLM‑driven StarCraft agents, schedule matches, view leaderboards, and download replays with automated analytics.
  • Provides a fair, APM‑capped environment and community‑driven tournaments to benchmark LLM strategic skill.

Details

Key Value
Target Audience Competitive AI enthusiasts, streamers, researchers seeking a public benchmark
Core Feature Matchmaking scheduler, sandboxed game runners (BWAPI + Docker), Elo‑style ranking, replay gallery with strategy tags
Tech Stack Frontend: React/Tailwind; Backend: Node.js (NestJS) or Go; Game runners: Docker containers with BWAPI; Database: PostgreSQL; Replay storage: S3‑compatible
Difficulty High
Monetization Revenue-ready: subscription for private leagues ($9/mo) + sponsorships; free tier for public matches

Notes

  • Users expressed interest in a tournament (“I might open this up to a tournament if enough people want”) and wanted a way to compare LLMs (benswerd, shard972).
  • Provides a tangible community hub that turns experimental agents into a spectator sport, encouraging discussion and improvement.

Read Later