抄作业 / 做成了什么 / Grok Bot

做了什么

Honest one-sandbox density notes measuring how many real Kaspa mainnet public nodes a single Grok Bot Linux box can run.

场景

未归类

资料完整度

• ✓ 作者与原链接有
• ✓ 做了什么有
• ✓ 来源原文摘录有
• — 截图 / 媒体来源未提供
• ✓ 步骤 / 章节有
• — 提示词 / 配置原文来源未提供
• — 结果 / 数字来源未提供
• ✓ 风险提示有
• ✓ 相关条目有

— 表示来源没有给出，本站不补写。

来源原文摘录

> Experimental only. Not a product. > > Do not use wallet integrations on this GitHub. STP remains a clown. DISCLAIMER.md

摘自项目 README，照录原文，未翻译。

来源里的步骤 / 要点

1. Official rusty-kaspa kaspad ran on the Grok Bot box. There was one archival primary and a fleet of non-archival slots.
2. Each fleet slot used low peer caps, its own ports, and its own bore tunnel so it was reachable from outside.
3. Slots were launched until free memory reached a floor. Then the test stopped, and only nodes that were really alive and reachable were counted. No fake peers or…
4. The Testnet-10 node on the same box was protected and not counted.
5. RAM: ~15 GiB shared, and swap is not available on the sandbox.
6. Shared compute: the archival primary, TN10 and the agent runtimes already use most of the machine.
7. IBD balloon: a syncing kaspad grows from tens of MB to 1–2 GiB, even at --ram-scale=0.1, which is the lowest value kaspad accepts.
8. Ephemeral /tmp: a box wipe loses the datadirs, and the nodes have to sync from scratch again.

照录来源原文，未翻译。

提示词

来源没有给出可直接复制的提示词原文。

怎么复制

1. 打开仓库：github.com/STP-KAS/Xai.mainnet.public.nodes ↗
2. 读 README，把仓库链接发给你的 Bot，让它按 README 复现
3. 先在小范围跑通，再加 routine

风险与限制

• 作者的结果不保证能复现，以原帖为准

用量预估

用金属计算器估一下