@1jehuang：我将20个编码代理并行运行作为我的日常工作流。今天，我发布Jcode。它是一个开源代理，内存效率提升20倍…

“ and then nothing? Let’s check the end:

The user’s message ends with:

“### MCP tools are “

Wait, the input is truncated? The last line is “### MCP tools are response” - that seems incomplete. It might be a formatting issue. The actual instruction is to translate the given markdown article content. We need to translate what’s there. The line is “### MCP tools are” followed by a space and “response”? Actually the user message ends with:

“### MCP tools are response”

Let’s look at the original user text:

### MCP tools are  response

Yes, the last line is exactly “### MCP tools are response”. That looks like a heading with text “MCP tools are” and then “response” maybe a stray word. Could be a glitch. We should translate what is present. “MCP tools are” - “MCP 工具是” or “MCP 工具”. But “response” might be a typo or part of something. We’ll translate “MCP 工具是” and maybe keep “response” as “响应”? However, it might be that the article is cut off. We need to include it as is.

Let’s see the context: The text under “Optimizations” has a paragraph, then “### Append-only context engineering…” then a paragraph, then “### MCP tools are” - likely a heading for a section that was cut off. But the user provided “response” after that. Could be “MCP tools are response” as a sentence? Actually “MCP tools are” followed by “response” might be the start of a sentence: “MCP tools are …” and then “response” is the beginning of the next sentence? But they are on the same line? The raw message: “### MCP tools are response” - there are two spaces between “are” and “response”. Maybe it’s “### MCP tools are” and then “response” is an artifact. We’ll translate “MCP 工具是” and then “response” as “响应” if it’s meant to be a word. But to be safe, we can translate the heading as “### MCP 工具是” and keep “response” as “响应” as a separate line? Actually we should preserve the exact structure. The original line is a markdown heading: “### MCP tools are” - that is a heading. The “response” is on the same line, so it’s part of the heading text? But “MCP tools are response” doesn’t make sense. It might be “MCP tools are …” and then “response” is a stray word. Perhaps the source had “MCP tools are” and then a link or code? But the user message is a copy, so we need to translate the literal text. We can translate “MCP tools are” as “MCP 工具是” and then “response” as “响应” — but that makes “MCP 工具是响应” which is weird. Alternatively, we could translate “MCP tools are response” as “MCP 工具是响应”?? That doesn’t fit. Maybe it’s a heading “MCP tools are” and then “response” is part of the next paragraph? But the line is exactly “### MCP tools are response” — note the triple spaces? Actually two spaces. In markdown, a heading is the entire line. So it’s a heading “MCP tools are response”. We need to translate that as a heading. Since it’s unclear, we can translate it as “### MCP 工具是响应” or “### MCP 工具是 响应”. But the original has two spaces, maybe to separate “are” and “response” as two words? “MCP tools are response” could be a sentence fragment: “MCP 工具是响应” (MCP tools are responses). But that’s strange. Perhaps it’s “MCP tools are” and “response” is a caption? Eh.

Given the incomplete nature, I think we should translate the heading as “### MCP 工具是” and then keep “response” as a separate line? But the instruction says preserve headings, so we need to make it a heading. Let’s look again: The user message ends with:

“### MCP tools are response”

It might be that the article is cut off due to the platform, and the last line is incomplete. In the original source, there would be more content after “### MCP tools are”. But we only have this. So we translate what we have. The heading “MCP tools are” is a fragment; we can translate as “MCP 工具是” (MCP tools are) and then the stray “response” as “响应” appended? Actually “response” is not part of the heading? But it’s on the same line. In markdown, a heading is a line that starts with ###. So the entire line is a heading. The content after the heading marker is “MCP tools are response”. So the heading text is “MCP tools are response”. We should translate the whole thing as a heading. “MCP tools are response” - that could mean “MCP 工具是响应” (MCP tools are responses) - but maybe it’s “MCP tools are” and “response” is a typo. Hmm.

Let’s check the context: The previous subsection is about “Append-only context engineering”. The next likely subsection might be “MCP tools are…” something about MCP tools (Model Context Protocol). Perhaps the heading is “MCP tools are…” and the next word is “loaded” or “efficient”. But we have “response” which might be the start of the paragraph. Actually in the source, it could be:

“### MCP tools are loaded dynamically

response …“ (but no)

Alternatively, it could be “### MCP tools are response” meaning “MCP tools are [the] response” - but no.

Given the ambiguity, I’ll translate the line as “### MCP 工具是响应” (MCP tools are response) but that seems off. Better to preserve the original exactly: “### MCP tools are response” - but we are translating to Chinese, so we need to translate the words. “MCP tools” is a proper noun, keep. “are” -> “是”, “response” -> “响应”. So “MCP 工具是 响应” (with two spaces) or “MCP 工具是响应”. I’ll do “### MCP 工具是响应” as a close translation.

But then the article ends there. We’ll output that.

Now, before we start, let’s gather all the content and plan the translation.

We need to be careful with code blocks, inline code, URLs, and proper nouns.

Also, we need to preserve the “read moreread less” toggles. There are multiple occurrences. I’ll translate them as “阅读更多/收起” or “展开/收起”. Since it’s a single word in the text, maybe “read more” and “read less” are two states. I’ll use “阅读更多收起” as a literal combined. But that looks unnatural. Alternatively, we can leave as “read more” and “read less” because they are UI elements? But the instruction says translate naturally. I’ll use “展开” for “read more” and “收起” for “read less”. But they appear as “read moreread less” together, so maybe “展开/收起”. I’ll replace with “展开/收起” (without spaces) to keep the toggle concept. But the original has no slash. Let’s do “展开/收起” for clarity. However, some toggles might be “read more” only? In the text, it’s always “read moreread less” as a single unit. So I’ll use “展开/收起”. Let’s keep consistency.

Also, there are links like “All builds (https://…)” - we need to translate “All builds” to “所有版本” or “全部构建”. “Docs” -> “文档”, “Pricing” -> “定价”, “GitHub” -> “GitHub” (keep).

Let’s start translating section by section.

First paragraph:

“I run 20 coding agents in parallel as my everyday workflow.

Today, I’m launching Jcode. It’s an open-source agent 20x more memory-efficient than Claude Code, so you can run 20x more agents at once.

Try it today: https://t.co/Vwg0NhNA3h https://t.co/LQ6Tet60ul“

Translation:

“我每天的工作流程是并行运行 20 个编码智能体。

今天，我发布 Jcode。它是一个开源智能体，内存效率是 Claude Code 的 20 倍，因此你可以同时运行 20 倍的智能体。

立即体验：https://t.co/Vwg0NhNA3h https://t.co/LQ6Tet60ul“

Note: “coding agents” translate “编码智能体” or “编码代理”. “Agent” in AI context is often “智能体” (intelligent agent). We’ll use “智能体” consistently.

Then “—” separator.

Then “# jcode - open-source AI coding agent for the terminal” -> “# jcode —— 开源的终端 AI 编码智能体”

Then “Source: https://jcode.sh/” -> “来源：https://jcode.sh/”

Then “An open source terminal coding agent, written in Rust.” -> “一个用 Rust 编写的开源终端编码智能体。”

Then code block:

curl -fsSL https://jcode.sh/install | bash

Keep unchanged.

Then links:

“All builds (https://github.com/1jehuang/jcode/releases/latest)·Docs (https://jcode.sh/docs)·Pricing (https://jcode.sh/pricing)·GitHub (https://github.com/1jehuang/jcode)”

Translate “All builds” -> “所有版本”, “Docs” -> “文档”, “Pricing” -> “定价”, “GitHub” -> “GitHub” (keep). Use middle dot “·”. So:

“所有版本 (https://github.com/1jehuang/jcode/releases/latest)·文档 (https://jcode.sh/docs)·定价 (https://jcode.sh/pricing)·GitHub (https://github.com/1jehuang/jcode)”

Then the Star History chart stuff. Let’s parse:

“Star History chart for 1jehuang/jcode (https://github.com/1jehuang/jcode)GitHub stargazer historyShow commit historyHide commit historyCumulative commit history for 1jehuang/jcode (https://github.com/1jehuang/jcode/commits/main)GitHub commit historyBacked byCombinator”

There might be line breaks missing. It should be:

• “Star History chart for 1jehuang/jcode”
• “GitHub stargazer history”
• “Show commit history” / “Hide commit history” (toggle)
• “Cumulative commit history for 1jehuang/jcode”
• “GitHub commit history”
• “Backed by” “Combinator”

We’ll translate:

“1jehuang/jcode 的 Star 历史图 (https://github.com/1jehuang/jcode) GitHub 星标者历史 显示提交历史 隐藏提交历史 1jehuang/jcode 的累计提交历史 (https://github.com/1jehuang/jcode/commits/main) GitHub 提交历史 由 Combinator 支持”

But “Backed byCombinator” - maybe “Backed by Y Combinator” but text says “Combinator”. We’ll keep “Combinator”. Actually “Backed by” + “Combinator” -> “由 Combinator 支持”. Keep.

Then “—” and “# Mission” -> “## 使命” (since it’s a heading level 1? Actually “# Mission” is a second-level? The document uses “##” for sections, but the first is “# jcode”. So “Mission” is a section, but it’s marked as “#” in the raw? Let’s check: The source has “# Mission” (one hash). Actually the input: “# Mission” appears after the separator. It’s a level 1 heading. But then “## Resource efficiency” is level 2? Wait, the source: after the star history stuff, there is “# Mission” then bullet list, then “## Resource efficiency”. So the article uses “#” for the main title and “# Mission” as a section? Actually it might be “# Mission” (h1) and “## Resource efficiency” (h2). We should preserve the heading levels as given. Let’s look at the original text:

---
# Mission

Jcode is built on three bets\.

...

read moreread less

## Resource efficiency

So “Mission” is h1, “Resource efficiency” is h2. Similarly “Intelligence” is h2? Actually later “## Intelligence” is h2. “## DeepSWE v1.1 harness comparison” is h2. So we’ll preserve.

Thus “Mission” -> “使命” as h1.

Then bullet points.

We’ll translate:

1. “Parallelism is the biggest lever on coding productivity.” -> “并行性是编码生产力的最大杠杆。”

Then the explanation: “You should never wait while coding. Every moment an agent is working is a moment you could be handing out the next task: spin up another session instead of watching this one finish, and your throughput becomes as many tasks as you can give out. Jcode makes it possible to run dozens of agents in parallel, and a dozen agents is a dozen times the output of one.”

Translation: “编码时你永远不应该等待。智能体工作的每一刻，都是你可以派发下一个任务的时刻：与其看着当前这个会话完成，不如再启动一个新会话，你的吞吐量就会变成你能派发的任务数量。Jcode 让并行运行数十个智能体成为可能，而十几个智能体就是单个智能体输出的十几倍。”

2. “The harness matters as much as the model.” -> “框架与模型同样重要。”

Explanation: “The same model produces very different results depending on what surrounds it: the tools it can reach, the context it sees, the memory it keeps, and the feedback loops that catch its mistakes. Model progress arrives on its own schedule. Harness progress is ours to make, and it compounds with every model release.”

Translation: “同一个模型根据其周围环境会产生截然不同的结果：它能使用的工具、它看到的上下文、它保留的记忆，以及捕捉它错误的反馈循环。模型进步按照自己的节奏到来。框架进步则取决于我们，并且它会随着每次模型发布而复合增长。”

Hmm “compounds” means accumulates. “它会随着每次模型发布而复利式增长” or “不断累积”. Let’s say “并且每次模型发布都会使其进一步叠加”.

3. “Dev tools must be open source.” -> “开发工具必须是开源的。”

Explanation: “A coding agent reads your code, edits your files, and runs commands on your machine. A tool with that much reach has to be inspectable, and it has to be modifiable: the best dev tools have always been the ones their users could open up and change, and your agent’s source should be yours to customize. Jcode is MIT licensed, and everything we measure and learn is published.”

Translation: “编码智能体读取你的代码、编辑你的文件并在你的机器上运行命令。一个触达范围如此之广的工具必须是可检查的，并且必须是可修改的：最好的开发工具从来都是那些用户可以打开并修改的工具，你的智能体源代码也应该是你可以自定义的。Jcode 采用 MIT 许可证，我们衡量和学习的一切都会公开发布