Auto-Arch Tournament vs tldr MCP Gateway

Auto-Arch Tournament is an autonomous research system where an AI agent iteratively proposes, implements, and validates microarchitectural improvements to a RISC-V CPU. Starting from a standard 5-stage pipeline, the loop runs hypotheses in parallel, each going through formal verification (53 symbolic checks), cycle-accurate simulation, multi-seed FPGA place-and-route, and CoreMark CRC validation. Only hypotheses that beat the current champion get merged; everything else gets discarded. Starting from 301 iterations/second, the system hit 577 iter/s (+92%) across 73 attempts in 9.8 hours — producing a design 26% faster and 40% smaller in LUTs than the baseline. The insight the author drives home is that the real innovation isn't the AI agent — it's the verifier. The orchestrator is hardcoded to prevent agents from manipulating their own evaluation gates, a simple but critical design constraint that turns a creative process into a trustworthy one. Without a rigorous verification harness, agent-driven optimization becomes a confidence trick. This is early but fascinating proof that AI-driven hardware design loops can produce commercially meaningful gains. The repo uses Claude Code or Codex as the coding agent, SystemVerilog for the RTL, and standard open-source EDA tooling (Yosys, nextpnr, Verilator). It's a compelling template for anyone building agentic optimization loops where correctness matters.

tldr is a local proxy that sits between your AI coding harness and upstream MCP servers, solving one of the most underappreciated problems in agentic workflows: context bloat from tool schema proliferation. When you connect GitHub MCP, filesystem MCP, and a few others, you can easily be sending 24,000+ tokens of tool schemas to the model before any work begins. Instead of passing all those schemas directly, tldr exposes exactly five wrapper tools to the model: search_tools, execute_plan, call_raw, inspect_tool, and get_result. The model learns which underlying tools exist on-demand through search_tools, then calls them through the proxy. GitHub MCP's 24,473-token schema surface compresses to 3,482 tokens — an 86% reduction. Output responses are further compressed through field stripping, a 4,096-token cap, and a 64KB byte limit. This is a genuinely practical solution for power users running multi-MCP setups who've noticed degraded performance as their tool count grows. The tradeoff is one extra hop of indirection, but the token savings pay for themselves in improved model attention and lower API costs.