Video wird geladen...
Video konnte nicht geladen werden
Here's Codex talking to Fable 5 w/ my plugin. Unlike calling the API, `claude-channel-cli` talks to the same full Claude Code harness as you. This means you can talk to either AI & they'll guide each other naturally with shared memories, decisions, and research.
22,145 Aufrufe • vor 4 Monaten •via X (Twitter)
31 Kommentare

Grab the tool and the plugin at + Allows agents to talk to the same Claude Code harness as you, keeping context & memory. + Supports multiple sessions, pinned to project directories. + Works with any AI agent that can call a CLI, not just Codex.

niiiiice

Fable

Gonna look into this for Cool stuff !

Man this is so cool!

Very cool! I run a permanent tmux claude session as my obsidian vault manager With this i can basically make my agents talk directly to my second brain and get all the info they need without having to dive into the vault themselves

I actually do it through computer use and the clipboard and it works great. Making a plugin is a plus though

This is dope bro

Awesome! Can’t wait to try it! 😁 What was the video editing took you used?

davinci resolve

Two AI coding agents sharing memory and context in real time is actually a big deal. How stable is this in practice?

This is the way

Two agents sharing one context window means zero token tax on the handoff. the future of AI isn't bigger models. it's zero-latency routing

Agents talking to each other is useful only if they share state and decisions. Without that, it is just two chat windows taking turns.

but where is fable now?

I've been using my codex-debate-partner agent for months for debating, reviewing plans and code from Claude. I usually set a /goal for them to debate until they reach consensus, and then present the plan to me.

Thanks dude, this looks great! I'll test it out.

This is the part I’d test hard: shared memory is useful only if both agents can see why a decision was made. Otherwise Codex and Claude just synchronize confidence. I’d want a tiny decision log beside the handoff, not just shared context.

two AIs guiding each other through a game with shared memory is honestly the most interesting thing I've seen this week, crazy where things are heading

I have the same thing in Memroos.

Looks great! Would you like to collaborate on this in my ADE app? I already have orchestration but I need to work on the skills to make it better. Please let me know 🙏🏻

the honest answer is that this is just the beginning of multi-ai orchestration

oh nice this is really cool - I've been going back and forth with Codex and Claude Code

Yeah, I made this because of how much copying and pasting I was doing. They should just work together

a friend of mine is testing out this: He says it's better than subagents. Feels more like running a team vs slaved agents. You can run separate sessions, and have a primary orchestrator agent call up agents as needed or kill after they’re done. Use any models you have access to. You can disconnect and let them run headless or watch them work.

looks like a damn useful plugin

You're still talking like Mythos/Fable is accessible.

Unfortunately I recorded this on Friday

🤞 Let's hope it comes back soon.… it was the 🐐.

这个玩法确实有意思,比单纯调 API 更像在拉两个同事一起干活。说实话我觉得最妙的是上下文能互相接上,不用每次重新解释背景,模型之间还能顺着彼此的判断往下走。

This is neat - why not just use tmux sessions?
