A reproducible VPS workbench for coding agents that keeps running when you close your laptop.
Codex and Claude can review each other’s diffs read-only, so the agent making changes isn’t the one judging them.
Terraform + Ansible, Tailscale-only SSH, backups, restore runbook, and optional private Firecrawl + MCP.
https://t.co/ZXry2eEn6V
@asaio87 Yes, and they are interchangeable. Just point in your agents.md to the .claude/skills folder or, vice versa, point in claude.md to .codex/skills. That way you can keep managing your skills from one folder.
@DumbleDax@SugarTrader77 I started experiencing issues using yfinance as well since earlier this year. I then switched to the yahooquery package and my data pulling issue has been resolved since then.
@SugarTrader77 Thanks for the feedback. I myself let it go pretty quickly, since the ATR-distance is very correlated with basic RSI. I did not quantify this but visually it’s pretty clear already. https://t.co/HMqHcyXAAq
@SugarTrader77 Cool! Personally I plot the ATR-multiple on a separate pane. Also I mark the mean + 2 sigma and mean + 3 sigma as calculated on the specific stock's history, to make the labeling of 'extension' truly stock-specific.
@Tolga_Inv@libernovaq@PradeepBonde@version You can use the regular screener to make a very broad watchlist and then scan on an indicator using PineScreener. Largest number of stocks I tested it for is 500.
@jfsrev@burritobandit04 Ty for everything you share. I use a TV-equivalent: screen the volatility adjusted, scaled and capped 8-32 EMA difference of the tickers versus SPY. Value is always between +20 (max. RS) and -20 (min. RS). Can show ATR-multiples distances and absolute strength at the same time.
@Inder_Mannster@TedHZhang Square root of 256, roughly the number of trading days in a year. His calculation represents the expected standard deviation of daily SPX return, and the ATR is proportional to this expected standard deviation, so wider intra-day ranges.