Anyone who's tried AI-assisted trading research has hit the same wall. The agent has no native access to your charts. You end up copy-pasting symbols, indicator values, screenshots, and Pine Script back and forth between TradingView and Claude or Cursor. The tools that try to fix this fall into two camps: route market data through a third-party API (added latency, added cost, their interpretation of your bars), or pollute your TradingView chart with helper indicators just so an agent can read them back. There's a third path that's more obvious in retrospect: drive your local TradingView Desktop through the Chrome DevTools Protocol that the app already exposes on port 9222 when you launch it with --remote-debugging-port=9222. The agent talks to the same TradingView you already use, reads what your chart actually shows, executes Pine Script through TradingView's own runtime. Same data, same indicators, same features, just with an agent in the loop. A project called tradesdontlie/tradingview-mcp star...

Every AI coding agent ships with the same problem: it knows syntax but not discipline. It can write a React component or debug a segfault, but it won't ask "did I verify this actually works?" before declaring victory. It won't split a 400-line diff into reviewable chunks. It won't check if the fix it's about to apply matches the root cause it claims to have found. I've spent the past six months fixing that. whetstone is a Claude Code plugin, and ai-skills is its portable counterpart for 35+ AI coding agents. Together they encode the engineering judgment that separates "code that compiles" from "code you'd ship." What's in the box The whetstone plugin ships 30 skills, 19 specialized agents, and 22 commands. Skills are compact instruction sets that fire based on what you're working on. Agents are purpose-built reviewers and researchers. Commands wire them into repeatable workflows. The ai-skills repo extracts just the skills into a format any agent can consume: Claude Code (what I use), Codex, Open...