Lightpath MCP server shows your AI agent exactly which code lines a click actually ran
Unknown-333 · reddit · 2026-11-06
A developer released Lightpath, an open-source MCP server that fixes a blind spot in agentic coding: the agent can read your code but can't see what actually ran.
- How it works: the agent names a button, Lightpath clicks it in its own Chrome instance, and returns every line executed in the browser and on the Node server — plus the lines that didn't run
- Example output: getSummary 14-28 ×1 not run: 16-27 — lines 16-27 never ran, which is the bug
- Works with Claude Code, Cursor, VS Code, Kiro, or any MCP client
- Source available on GitHub
More from coding & agent
- Team livestreams open RL training of LoRA harness adapters for hermes, pi and opencode — yb2698 · 2026-11-06
- Workers IO launches to verify AI-written code with millions of fault-injected production simulations — julianweisser · 2026-11-06
- Law professor says Claude deleted a year of lecture notes at lunch — JeremyNguyenPhD · 2026-11-06
- 7 techniques for building cost-efficient AI pipelines, from model routing to cost-per-task — goyalshaliniuk · 2026-11-06
- Claude Code leaked users' emails via User-Agent header during web searches — NousResearch · 2026-11-06
- Harness-S rescues failed RL rollouts with sibling harnesses for generalizable agents — SJUT1 · 2026-11-06