Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Message your other Claude Code sessions

code.claude.com

Post reply on HN
Print view View original

61–70 of 76 posts

Re: Message your other Claude Code sessions

#61

Post by Yokolos »

eigenblake wrote: post #10

Like some others, I also built this myself. Overly simply, with tmux, a memory tree, and handoff files and an orchestrator. And yet for how simple it was, it was so effective at minimizing the amount of duplicate context. It's like having shared specialist subagents who source and derive important shared knowledge from separate threads. It's useful because some skills just take too much of a token penalty to invoke a…

Have you maybe documented this anywhere or do you have suggestions for repos to look at that do this? This sounds interesting and I'd like to look into this myself

Re: Message your other Claude Code sessions

#62

Post by fragmede »

nizbit wrote: post #58
zurfer wrote: post #42

This is obviously cool and useful so kudos, but wow security researchers have to throw their hands up all the time. Now we open another attack surface where you can ask a remote agent to do things by default. There was a time when you call this a Remote Code Execution vuln. It's of course a feature here.

Making calls across a trust boundary, what could go wrong?

It's Claude code talking to another Claude code. They're both already leaking your data to Anthropic in the first place.

Re: Message your other Claude Code sessions

#63

Post by fragmede »

SyneRyder wrote: post #48
zhoujianfu wrote: post #13

Earlier quoted context omitted.

How do you know when they’re done (or need your attention)? Is that a standard “api”/hook? I want that!

Claude Code has hooks: https://code.claude.com/docs/en/hooks-guide In your Claude settings.json file, you can configure a program to run when it needs input. In my case, I wrote a small exe that sends a notification via Pushover. You can configure Claude Code to run the hook on "idle_action" (the agent is done), "question" (the agent needs to ask you something), or on a "PermissionRequest". There are other triggers p…

> In your Claude settings.json file, you can configure a program to run when it needs input.

I asked Claude to tell me (using TTS) what it just finished doing instead of "configuring a program to run".

Re: Message your other Claude Code sessions

#64

Post by fragmede »

The bigger question is, how does this affect the mythical man month? Nine women can't make a baby in one month is the axiom, but if you have 14,000 women, chances are, one of them is 8 months pregnant and you'll get a baby in one month. If you want one tomorrow, go steal one from the ICU. Coordination between humans is difficult, and this meeting could have been an email. AI agents can also work together more efficiently than humans can. None of us is as dumb as all of us, unless it's an AI. Does adding an additional AI agent to a project make it ship faster or slower?

The world's about to find out!

Re: Message your other Claude Code sessions

#66

Post by mceachen »

simonw wrote: post #19
Alifatisk wrote: post #3

I wish there was a clean way to compact the conversation into a prompt with all necessary context for a new fresh conversation.

I'm fed up with compaction. I want my agent to get compacted but also retain full access to the prior conversation via search and tool calls - I want it to know "the requirements for X were discussed in detail previously in conversation C51E31CE-C985-4633-A749-DCC9805A7FEB" and have a tool that lets it dispatch a subagent to find those details again. Do any of the coding agents have this already?

Claude is quite good at searching both its own and codex session transcripts, and if asked, can spawn tmux/konsole/whatever to resume those sessions to be interactive if needed.

My system prompt adds TPPs and TLDRs for task/feature planning, and shared memories/documentation management. I find them invaluable.

https://photostructure.com/coding/claude-code-tpp/

Re: Message your other Claude Code sessions

#67

Post by aqme28 »

Alifatisk wrote: post #55
aqme28 wrote: post #37

Earlier quoted context omitted.

Just make a new slash command with those instructions as the prompt.

Seems like that's the only way at the moment. Create a good general handoff prompt and turn it into a SKILL.

Not sure what other way you’d even want. It’s super easy.

Re: Message your other Claude Code sessions

#68

Post by 3371 »

simonw wrote: post #19
Alifatisk wrote: post #3

I wish there was a clean way to compact the conversation into a prompt with all necessary context for a new fresh conversation.

I'm fed up with compaction. I want my agent to get compacted but also retain full access to the prior conversation via search and tool calls - I want it to know "the requirements for X were discussed in detail previously in conversation C51E31CE-C985-4633-A749-DCC9805A7FEB" and have a tool that lets it dispatch a subagent to find those details again. Do any of the coding agents have this already?

In my own framework (just a skill), every unit of works produce a markdown document. This provides rich context over past plan/interview/implementation/etc., this + git is more than enough to not think about context and memory.

Re: Message your other Claude Code sessions

#69

Post by ulrikrasmussen »

zhoujianfu wrote: post #13
andai wrote: post #4

I was reading about the HF hack and one of the first thing the GPT swarm did was build a messaging system for themselves. This reminded me of that. (Also of how weirded out I was by Moltbook when it came out. Time flies!) Unrelated: when my Claude/Codex finishes cooking (or needs my attention), it pings a local HTTP endpoint that plays a frog sound on my laptop. I found this massively boosts quality of life.

How do you know when they’re done (or need your attention)? Is that a standard “api”/hook? I want that!

I have a shell script that sends a message to me on Slack, and I have instructions in CLAUDE.md that it should be invoked when a long task finishes or if my attention is needed. The Slack app API key is one way messaging, so it doesn't allow Claude to exfiltrate anything.

Re: Message your other Claude Code sessions

#70

Post by cadamsdotcom »

Alifatisk wrote: post #52
anonym00se1 wrote: post #29

Earlier quoted context omitted.

Create your own protocol. I created a "wind down session" protocol my agents use that takes detailed notes in a "next_session_prompt.md" file that covers what was done this session, what is still open, and where they need to pick up the next session. You can refine the protocol as you realize what's working and what isn't. I've been using that for months and it rarely drops important things now.

You mean like a SKILL?

English instructions in a text file plus an instruction you give, to read the file and do what it says.

Protocol is probably not the precise word, but even so, a skill as currently defined covers way more ground none of which is needed here

Post reply on HN