Hacker News: parweb

Auto-compaction felt fine. The invoice didn't

parweb — Mon, 02 Feb 2026 00:42:31 +0000

I let Claude Code context accumulate and burned $2,000+ in tokens — without noticing any quality drop

I made a costly mistake with Claude Code: I let context accumulate across long sessions and relied on auto-compaction. What surprised me is that the output quality stayed good. I was happy with the results. There was no obvious “something is wrong” signal. The only thing that went wrong was the bill.

What happened - I kept iterating in the same long-lived context. - Auto-compaction kicked in when needed. - Sonnet was used by default most of the time. - From a UX perspective, everything felt fine.

Why this is dangerous - There’s no clear quality degradation to warn you. - Token usage grows invisibly in the background. - Auto-compaction itself consumes a lot of tokens. - You only realize something is wrong when you look at the invoice.

Root causes - Long-lived context feels convenient, so you don’t reset it. - Tooling doesn’t surface cost as a first-class signal. - The mental model (“LLMs can remember my project”) is seductive but expensive. - Defaulting to a large model makes the problem much worse.

What I changed - One task = one fresh context. - Externalized memory: project state lives in `context.md` / `decisions.md`, not in prompt history. - Default to a smaller model; large models only for design/architecture. - Diff-only outputs: no full file re-dumps. - Disabled auto-compaction; summaries now live in docs. - Added cost visibility: token counters and budget caps.

Takeaway This isn’t about output quality degrading. It’s about cost scaling quietly without feedback. LLMs make it too easy to accumulate invisible technical debt in tokens.

If tooling doesn’t make cost visible, people will keep doing this.

Comments URL: https://news.ycombinator.com/item?id=46851021

Points: 2

# Comments: 0

Show HN: Dynamic Prisma wrapper for Jotai with full TypeScript and pagination

parweb — Thu, 05 Jun 2025 12:47:18 +0000

This library creates a bridge between Prisma and Jotai by: - Using *dynamic proxies* to intercept model access - Leveraging *generated Prisma types* (`Prisma.TypeMap`, `Prisma.ModelName`) - Creating *cached atom families* for optimal performance - Providing *advanced TypeScript inference* for conditional return types

Comments URL: https://news.ycombinator.com/item?id=44191146

Points: 2

# Comments: 0

New comment by parweb in "[dead]"

parweb — Sun, 05 Jan 2025 00:25:29 +0000

alright just come and hangout ^^

New comment by parweb in "[dead]"

parweb — Sun, 05 Jan 2025 00:23:45 +0000

Thank you for asking! Here’s a clearer breakdown:

Responsibilities

Recruitment: Each Elite Member will recruit 10 new members, who will then recruit others, creating a scalable network. Leadership: You'll guide your recruits, ensuring they understand their role in building and engaging with the SocialOps community. Collaboration: You'll contribute ideas and strategies to help the team create and execute viral trends. Time Commitment

Weekly Time: Around 2-3 hours a week to recruit, communicate with your team, and collaborate with other Elite Members. Goals

Recruit your team of 10 within a defined timeframe. Help shape and execute our first major viral trend strategy. Support Provided

Access to resources and tools to help with recruiting and planning. Regular updates and strategy sessions with the core SocialOps team. Let me know if this makes sense or if you have further questions! @socialops