d4rkp4ttern

7 ms·

1. 61.▲
   by d4rkp4ttern 5mo ago

   I try to counteract this "comprehension-debt" by having the code-agent quiz me, Socratic-style, about the core problems, possible solutions, why certain approaches wouldn't work, and why the specific approach was implemented.
2. 62.▲
   by d4rkp4ttern 5mo ago

   A huge factor I don’t see mentioned often enough, is the rapid increase of AI-coding in a language unknown to the dev.
3. 63.▲
   by d4rkp4ttern 5mo ago

   Most of the narrative is about how AI is writing all/most code, but I’d wager that the fraction of human reviewed code is approaching zero far faster than anyone is realizing or willing to admit.
4. 64.▲
   by d4rkp4ttern 5mo ago

   Another one like this - https://sinceyouarrived.world/taken
5. 65.▲
   by d4rkp4ttern 5mo ago

   This should probably be merged with this: https://news.ycombinator.com/item?id=48177107
6. 66.▲
   by d4rkp4ttern 5mo ago

   I agree with this (and the earlier comment about perceived expertise vs actual expertise), and I think it goes beyond hiring managers. The core demoralizing fact is that when people perceive that AI can give results at least as good as huma
7. 67.▲
   by d4rkp4ttern 5mo ago

   I can see where it's coming from. Putting it starkly, at a high level, the broad effect of AI is this: devaluation of expertise, whether in coding, or drawing, or music composition, or writing, or translation, or so many other
8. 68.▲
   by d4rkp4ttern 5mo ago

   In the context of coding agents, one thing I find surprisingly useful is this - rather than have the code-agent explain to me what it did and why, I have it quiz me Socratic-style: it presents a scenario or problem, and asks me why a certai
9. 69.▲
   by d4rkp4ttern 6mo ago

   I guess you mean for STT. For my usecase of talking to AI's or coding agents, pure STT accuracy is less important than transcription speed. Transcription needs to be near-instant, and accuracy "good enough" so that the AI&#x2
10. 70.▲
    by d4rkp4ttern 6mo ago

    This was on HN 7 months ago: https://news.ycombinator.com/item?id=45114245 Every time a STT/TTS model is posted I wonder if it will change my current workflow on MacOS, which is: STT with Parakeet-V3 via Hex [1] app fo
11. 71.▲
    by d4rkp4ttern 6mo ago

    My regular workflow is to run code agents in Tmux panes and often I have Claude Code consult/collaborate with Codex, using my tmux-cli [1] tool, which is a wrapper around Tmux that provides good defaults (delay etc) for robust sending
12. 72.▲
    by d4rkp4ttern 6mo ago

    I now often have CC make technical/architecture diagrams with tikz, the results look much better than mermaid but still requires multiple iterations to fix bad arrows, bad layouts etc. Diagrams are still far from solved. We need a good
13. 73.▲
    by d4rkp4ttern 7mo ago

    you can set it in the .omlx/settings.json - ask a code-agent to figure it out by pointing it at the omlx repo
14. 74.▲
    by d4rkp4ttern 7mo ago

    At least for the Gemma4-26B-A4B, Token-gen speed with OMLX is far worse on my M1 Max 64GB Macbook, compared to llama-server: Quick benchmark on M1 Max 64GB, Gemma 4 26B-A4B (MoE), comparing matched dynamic 4-bit quants. Workload was C
15. 75.▲
    by d4rkp4ttern 7mo ago

    Correct, no issues because since at least a few months, llama.cpp/server exposes an Anthropic messages API at v1/messages, in addition to the OpenAI-compatible API at v1/chat/completions. Claude Code uses the former.
16. 76.▲
    by d4rkp4ttern 7mo ago

    You can use llama.cpp server directly to serve local LLMs and use them in Claude Code or other CLI agents. I’ve collected full setup instructions for Gemma4 and other recent open-weight LLMs here, tested on my M1 Max 64 GB MacBook: https:
17. 77.▲
    by d4rkp4ttern 7mo ago

    Thanks, fixed
18. 78.▲
    by d4rkp4ttern 7mo ago

    For token-generation speed, a challenging test is to see how it performs in a code-agent harness like Claude Code, which has anywhere between 15-40K tokens from the system prompt itself (+ tools/skills etc). Here the 26B-A4B variant i
19. 79.▲
    by d4rkp4ttern 7mo ago

    I dug into this more. It's disabled by default, and it's a cost/token-usage optimization. The logic is: 1. Anthropic's API has a server-side prompt cache with a 1-hour TTL 2. When you're actively using a se
20. 80.▲
    by d4rkp4ttern 7mo ago

    For me one of the most interesting aspects is how compaction works. It turns out compaction still preserves the full original pre-compaction conversation in the session jsonl file, and those are marked as "not to be sent to the API&quo
21. 81.▲
    by d4rkp4ttern 7mo ago

    that frustration regex is missing "idiot", which is the most common frustration word I use with code-agents
22. 82.▲
    by d4rkp4ttern 7mo ago

    Another option I use open is to ask the code-agent to make a diagram using Tikz (as a .tex file), which can then be converted to pdf/png. But in general AI-diagramming is still unsolved; needs several iterations to get rid of wonky&#x
23. 83.▲
    by d4rkp4ttern 7mo ago

    Agree. For what it’s worth, in interviews Cherny (Claude Code creator) and Steinberger (OpenClaw creator) say they keep things simple and use none of the workflow frameworks. The latter even said he doesn’t even use plan mode, but I find th
24. 84.▲
    by d4rkp4ttern 7mo ago

    Thanks - I use worktrees and direnv but never thought about this ennrc trick to auto-share .env across worktrees.
25. 85.▲
    by d4rkp4ttern 7mo ago

    I see lots of discussions about humans no longer writing code but the elephant in the room is the rapid extinction of human-review of AI-made code. I expect this will lead to a massive hangover. In the meantime we try to mitigate this by e
26. 86.▲
    by d4rkp4ttern 7mo ago

    For every new interesting open model I try to test PP (prompt processing) and TG (token gen) speeds via llama-cpp/server in Claude Code (which can have at least 15-30K tokens context due system prompt and tools etc), on my good old M1
27. 87.▲
    by d4rkp4ttern 8mo ago

    I had exactly this problem and didn’t see anything good out there (Claude —resume only searches session names and auto-created titles) so I got a tool built that uses a Rust/Tantivy full text search index. It’s part of the aichat comma
28. 88.▲
    by d4rkp4ttern 8mo ago

    Yes that one is great too
29. 89.▲
    by d4rkp4ttern 8mo ago

    I’m intrigued by their recent CLI release as well. I’ll have to check out the markdown edit support too, thanks
30. 90.▲
    by d4rkp4ttern 8mo ago

    Sesame was the best full-duplex voice demo I ever came across, wonder what is up with them now https://app.sesame.com/