Hacker News: frog437https://news.ycombinator.com/user?id=frog437Hacker News RSShttps://hnrss.org/hnrss v2.1.1Thu, 09 Apr 2026 12:35:49 +0000<![CDATA[New comment by frog437 in "Ask HN: What are you working on? (April 2026) (Non AI)"]]>It's a CLI tool. It plays well with AI but it works great without it.

]]>Tue, 07 Apr 2026 18:30:56 +0000https://news.ycombinator.com/item?id=47679426frog437https://news.ycombinator.com/item?id=47679426https://news.ycombinator.com/item?id=47679426<![CDATA[New comment by frog437 in "Ask HN: What are you working on? (April 2026) (Non AI)"]]>Understand an entire codebase instantly : www.ix-infra.com

]]>Tue, 07 Apr 2026 18:30:07 +0000https://news.ycombinator.com/item?id=47679411frog437https://news.ycombinator.com/item?id=47679411https://news.ycombinator.com/item?id=47679411<![CDATA[New comment by frog437 in "Issue: Claude Code is unusable for complex engineering tasks with Feb updates"]]>This is crazy..

tokensSaved = naiveTokens - actualTokens

  - naiveTokens = 19.4M — what ix estimates it would have cost to answer your queries without graph intelligence (i.e., dumping full files/directories into context)                                    
  - actualTokens = 4.7M — what ix's targeted, graph-aware responses actually used
  - tokensSaved = 14.7M — the difference

]]>Tue, 07 Apr 2026 04:04:14 +0000https://news.ycombinator.com/item?id=47670632frog437https://news.ycombinator.com/item?id=47670632https://news.ycombinator.com/item?id=47670632<![CDATA[New comment by frog437 in "Issue: Claude Code is unusable for complex engineering tasks with Feb updates"]]>depending on how large your codebase is, hopefully not. At this point use something like the IX plugin to ingest codebase and track context, rather than from the LLM itself.

]]>Tue, 07 Apr 2026 01:17:24 +0000https://news.ycombinator.com/item?id=47669568frog437https://news.ycombinator.com/item?id=47669568https://news.ycombinator.com/item?id=47669568