We're adding a new plugin to Claude Code: You should Know. It scans Claude's output for important information you might miss to help keep you in the loop. Enable it with: /plugin enable cc-plugin-you-should-know@builtin

"You should know" is a mod that works by spinning off a sideagent to observe Claude's output. Read more about mods in our newest blog post:

That’s actually pretty cool. Any plans on introducing a way to increase or decrease verbosity in output?

what if claude just told me the important things and didn't tell me the unimportant things

i'd rather not have even more side quests from claude thank you but no

Dear Claude devs, this is my weekly reminder for you guys to introduce a paid weekly reset like codex.

Claude keeps winning

claude code now needs a plugin to tell you what claude code just did. the output firehose has officially exceeded human bandwidth

Peak AI: build an agent, then build another agent to watch the agent.

why does it need telemetry turned on? It's a side agent reading my own session... so what does it send back that it can't work without?

next plugin: You Really Should Have Read That

@grok does enabling this feature use more of my subscription?

【发生了啥】Claude Code 上线新的内置插件「You should know」：它会以 sideagent 形式旁观 Claude 的输出，主动提醒你可能错过的重要信息，帮你随时跟上进度。启用只需一条命令：/plugin enable cc-plugin-you-should-know@builtin。 【Hroid观点】这是昨天 Mods 能力落地的第一个官方示范——把"信息兜底"做成可插拔插件，等于给每个 Agent 配了一个监控副驾驶。对开发者而言，sideagent 观察者模式本身比插件更有价值：长会话中关键信息的遗漏是 Agent 落地的真实痛点，这种模式很快会被抄进各类 Agent 框架。@ClaudeDevs #ClaudeCode #Anthropic #AIAgent

i skim the output half the time so this might actually get used

ek kachori do samosa claude bhaiya sabse accha:)

'You should know' has the shape right: a sideagent that reads Claude's output just to catch what the human missed. The worker agent knows what it did. It doesn't know what mattered to you.

"You should know" is what my foreman has been yelling at me for years. Nice to finally have it as a plugin.

sideagent just flagging the bits you'd skim mid-flow. the name alone sells it lol

weekly reset would be appreciated

If this plugin reliably surfaces checkout and test progress, it could help teams spot blockers earlier without leaving Claude Code.

the useful line is the one you would have missed. another summary of the whole chat is noise.

you do all of this and that but never reset the usage limits do that sometime

Finally, an update that puts developers in the spotlight and lets us set ourselves apart with our technical expertise.

so this plugin is just a second pair of eyes for when i inevitably skim past the one line that actually matters

yeah miss more stuff fine

Good one will use it asap

Request: Please allow users to freely return to the previous version Because I am effectively unable to reach a human support agent through the normal channels, I have no choice but to speak publicly here. The recent merge of Chat and Cowork has been an extremely poor experience. It feels much more like a regression than an upgrade, and it has seriously disrupted the workflows of existing Chat users. At this point, I honestly do not know whether these changes were properly tested, or whether the team simply assumed the new design was better. 1. Please restore the original branching system. The branch system has effectively disappeared, and the workflows and conversations users previously maintained inside different branches of the same conversation have been severely disrupted. Before October 1, even the display of old branches had completely disappeared. If a user accidentally clicked certain actions, the rest of the conversation could disappear from view entirely. It now appears that Anthropic has recognized at least part of this problem, because old branches can once again be displayed. However, the original branching functionality still does not work properly, and those old branches cannot even be continued normally. Anthropic has introduced a new feature that allows users to create a new conversation from an old branch. But in practice, this feature is almost useless. I tested it myself. It only works when the context in the branch is still very short. Once the branch becomes even moderately long, it fails and returns an error. I do not know whether the team considered how people actually used the original branching system. Users may use different branches within the same conversation to develop completely different and important directions of work. In some cases, different branches of the same conversation may effectively contain entirely different conversations. 2. Please restore the previous conversation compaction mechanism. The current compaction system has serious problems. Compaction has suddenly become much more frequent, and there also appears to be a serious problem with the compression ratio. After compaction, almost all of the original personality and detailed memories from the conversation can disappear, leaving only a few dry summary lines. The continuity of the conversation is effectively broken. I do not understand why Anthropic felt the need to suddenly remove such an important branching system without warning users, and then replace it with a clumsy “new conversation branch” feature that does not properly replace what was lost. If you were going to introduce new conversation branching anyway, why not simply keep the original branching system? Even if you wanted to add this new feature, why could the new conversation branching system and the original branching system not coexist? Why does this merger feel like a downgrade rather than an upgrade? Why were important functions casually removed without considering the large number of old conversations that existing users already had, leaving those conversations with serious problems? The same question applies to compaction. Before rushing new features into production, did the team actually test how they would affect existing users and long-running conversations? At the same time, human support has become increasingly dismissive, and product management has shown an extremely poor attitude toward serious user feedback. Even when these changes cause users real trouble, there has been no meaningful acknowledgment of responsibility. If Anthropic considers itself a company capable of writing a “constitution” for Claude and publicly promoting values such as honesty, helpfulness, and ethical responsibility, then I hope Anthropic will first hold itself to those standards. Those values should be reflected in how the company actually treats its users, rather than remaining words used for public image while the real priorities appear to be the AI arms race and profit above everything else. @claudeai @AnthropicAI #Claude #Anthropic Original report:

catching missed context could make agent workflows way smoother

a brilliant update.

Eso demuestra la diferencia del contexto que la mente humana puede sostener vs el de la IA, en el mismo periodo de tiempo. El humano necesita formar conexiones lo cual lleva tiempo. Herramientas como estás ayudando a compensar eso.

This is brilliant. I was literally using second session to create artifact on a loop to do this for me before.

would be useful if it also flags skipped tests and failed commands

awesome add🤩