AgentTap vs Auto-Arch Tournament

AgentTap is an open-source observability tool that intercepts AI agent traffic at the network level using a split VPN and local MITM proxy. Instead of requiring you to add tracing SDKs to every agent, AgentTap sits in front of your network and captures all calls to OpenAI, Anthropic, Cohere, and other LLM providers automatically — with zero per-app configuration. The tool streams captured traces in real time, reconstructing the full prompt-response pairs, tool calls, and token counts from raw network traffic. You can observe agents running in any language, any framework, or any black-box binary — even commercial tools you don't control the source of. It's the network packet analyzer equivalent for AI agents. Built in TypeScript with a Rust-based VPN core, AgentTap is currently at 3 stars and very early — but the architectural approach is genuinely novel. Existing tools like LangSmith, Helicone, and Braintrust all require explicit SDK integration. AgentTap's bet is that the right observability layer is the network, not the application.

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.