What 1,000+ Harness Experiments Taught Me About Self-Improving Agents (opens in new tab)

henrypan.com··Hacker News, Hacker News·Cited by 1 article·Covers: MCP is an open protocol that standardizes how apps provide context to LLMs, karpathy/autoresearch: AI agents running research on single-GPU nanochat training automatically +3 more·Open original

Project Repository: So I recently wanted to see whether an AI agent could self-improve a harness to solve terminal bench tasks. To align on the definitions, “harness” means the system (e.g. Claude Code, Codex, ChatGPT web interface etc…) wrapping around the model (e.g. GPT 5.5, Claude Opus 4.7 etc…)...

Read the original article

Sign in to keep reading the full article.

Sign Up Log In

Cited by 1 article

#057 - Anthropic hits $30B ARR, DuckDuckGo jumps 28%, and one Host header bypasses every FastAPI MCP

indiehacker.news·