Hacker News: _alphageek

New comment by _alphageek in "Reverse-engineering a supply chain attack delivered via fake Web3 job interview"

_alphageek — Wed, 22 Apr 2026 14:33:57 +0000

These attacks evolve each year. Initial ones were just obfuscated code inside some utilities jest/tailwind config, most sophisticated I have seen obfuscated code loaded on chain. So you can not find any trace in the packages, but when you start/install it loads transaction info - decode and boom - you are hacked. So the safest way not to run it or run in isolated docker environment.

New comment by _alphageek in "AI fact-checker with guardrail classifier and MCP server"

_alphageek — Wed, 22 Apr 2026 14:30:07 +0000

How it is different from perplexity prompt?

New comment by _alphageek in "Show HN: Minimal visual chess openings explorer"

_alphageek — Wed, 22 Apr 2026 10:49:57 +0000

Really nice.

If I were you I would also develop interactive opening tree as an embeddable widget for chess blogs, coaches, and content creators. Chess.com is static one.

Also if you can get api/history of user plays - you can help them with analyze to get better openers and become better players.

New comment by _alphageek in "How to make a fast dynamic language interpreter"

_alphageek — Tue, 21 Apr 2026 21:40:56 +0000

Good writeup. The Arguments arc (#7→#13) hits close — did basically the same dance for an async step evaluator in Rust a while back. Went all in on Cow<'_, Input> assuming borrow-in-the-common-case would earn its keep. Microbenches looked great. Real workload: the Cow discriminant plus lifetime gunk bled into every combinator past the first await, inlining fell off a cliff, the whole point of Cow evaporated. Ripped it out for NoInput / OneInput / MultiInput(Vec) at the evaluator boundary — same split as your ZeroArguments / OneArgument / TwoArguments, just arrived at the ugly way. One thing I keep wondering: have you stacked arity specialization with type specialization on the native path? Binary style, drops the isInt probe altogether. Guessing the code size math didn't work out, or ICs are already soaking up whatever's hot on the object side so the native fast paths don't matter much. Which one?

New comment by _alphageek in "Kimi vendor verifier – verify accuracy of inference providers"

_alphageek — Tue, 21 Apr 2026 21:38:47 +0000

Once vendors optimize for 6 KVV benchmarks, they measure compliance with KVV, not model fidelity. Is there rotation strategy in place?