Show HN: Mysti – Claude, Codex, and Gemini debate your code, then synthesize

2025-12-2313:18216178github.com

AI coding dream team of agents for VS Code. Claude Code + openai Codex collaborate in brainstorm mode, debate solutions, and synthesize the best approach for your code. - DeepMyst/Mysti

Show article

Your AI Coding team for VSCode
Use Claude Code, Codex, or Gemini — or combine any two in Brainstorm Mode and never hit bottlenecks
Wisdom of the crowd where the collective intelligence of several agents outperforms a single one.

Providers • Brainstorm • Features • Quick Start • Config

From VS Code: Press Ctrl+P (Cmd+P on Mac), then paste:

ext install DeepMyst.mysti

Or install from the VS Code Marketplace

Mysti works with the AI coding tools you already have. No extra subscriptions needed.

Provider	Best For
Claude Code	Deep reasoning, complex refactoring, thorough analysis
Codex	Quick iterations, familiar OpenAI style
Gemini	Fast responses, Google ecosystem integration
Brainstorm Mode	Any two AIs collaborate and debate solutions

Switch providers with one click. No lock-in.

vs Copilot/Cursor	Mysti Advantage
Single AI	Multi-agent brainstorming — two AIs collaborate
Locked to one provider	Use any CLI — Claude, Codex, or Gemini
Black box	Full permission control — read-only to full-access
Generic responses	16 personas — architect, debugger, security expert...

Beautiful, modern chat interface with syntax highlighting, markdown support, and mermaid diagrams

Want a second opinion? Enable Brainstorm Mode and let two AI agents tackle your problem together. Choose any 2 of 3 agents (Claude, Codex, or Gemini) from the settings panel.

Claude Code (Anthropic), Codex (OpenAI), and Gemini (Google) have different training, different strengths, and different blind spots. When any two work together:

Each AI catches edge cases the other might miss
Different perspectives lead to more robust solutions
Together they debate, challenge each other, and synthesize the best solution

It's like having a senior dev and a tech lead review your code—except they actually discuss it first.

Configure which two agents collaborate in the Settings Panel:

Combination	Best For
Claude + Codex	Deep analysis meets rapid iteration
Claude + Gemini	Thorough reasoning with fast validation
Codex + Gemini	Quick iterations with Google ecosystem knowledge

Your Request
     |
     v
+-----------+-----------+
|  Agent 1  |  Agent 2  |
| analyzes  | analyzes  |
+-----+-----+-----+-----+
      |           |
      v           v
+---------------------------+
|   Discussion (Full Mode)  |
| Agents review each other's|
| solutions and debate      |
+-----------+---------------+
            |
            v
+---------------------------+
|        Synthesis          |
| Best ideas combined into  |
| one refined solution      |
+---------------------------+

Quick Mode	Full Mode
Direct synthesis	Agents discuss first
Both agents respond, then merge	Each AI critiques the other's solution
Faster results	More thorough analysis
Good for simple tasks	Best for complex architecture decisions

When the AI presents multiple implementation approaches, Mysti automatically detects them and lets you choose your preferred path.

Requires at least 2 CLI tools installed. See Requirements.

Shape how your AI thinks. Select from specialized personas that change the AI's approach to your problems.

Persona	Focus
Architect	System design, scalability, clean structure
Debugger	Root cause analysis, bug fixing
Security-Minded	Vulnerabilities, threat modeling
Performance Tuner	Optimization, profiling, latency
Prototyper	Quick iteration, PoCs
Refactorer	Code quality, maintainability
+ 10 more...	Full-Stack, DevOps, Mentor, Designer...

Select personas directly from the toolbar without opening panels.

Mysti automatically suggests relevant personas and actions based on your message.

Never lose your work. All conversations are saved and easily accessible.

Get started fast with one-click actions for common tasks.

Fine-tune every aspect of Mysti including token budgets, access levels, and brainstorm mode.

Already paying for Claude, ChatGPT, or Gemini? You're ready to go.

Mysti works with your existing subscriptions—no additional costs!

CLI Tool	Subscription	Install
Claude Code (recommended)	Anthropic API or Claude Pro/Max	`npm install -g @anthropic-ai/claude-code`
Codex CLI	OpenAI API	Follow OpenAI's installation guide
Gemini CLI	Google AI API or Gemini Advanced	`npm install -g @google/gemini-cli`

You only need one CLI to get started. Install any two to unlock Brainstorm Mode.

Option A: Press Ctrl+P (Cmd+P on Mac), paste and run:

ext install DeepMyst.mysti

Option B: Install from VS Code Marketplace

# Claude Code (recommended)
npm install -g @anthropic-ai/claude-code
claude auth login # Or Gemini CLI
npm install -g @google/gemini-cli
gemini auth login

For Brainstorm Mode, install any two CLI tools.

Click the Mysti icon in the Activity Bar, or
Press Ctrl+Shift+M (Cmd+Shift+M on Mac)

Type your request and let the AI assist you!

Mix and match behavioral modifiers:

Concise - Clear, brief communication
Test-Driven - Tests alongside code
Auto-Commit - Incremental commits
First Principles - Fundamental reasoning
Scope Discipline - Stay focused on the task
And 7 more...

Stay in control of what the AI can do:

Read-only - AI can only read, never modify
Ask-permission - Approve each file change
Full-access - Let the AI work autonomously

{ "mysti.defaultProvider": "claude-code", "mysti.brainstorm.agents": ["claude-code", "google-gemini"], "mysti.brainstorm.discussionMode": "full", "mysti.accessLevel": "ask-permission"
}

Setting	Default	Description
`mysti.defaultProvider`	`claude-code`	Primary AI provider (`claude-code`, `openai-codex`, `google-gemini`)
`mysti.brainstorm.agents`	`["claude-code", "openai-codex"]`	Which 2 agents to use in brainstorm mode
`mysti.brainstorm.discussionMode`	`quick`	`quick` or `full`
`mysti.accessLevel`	`ask-permission`	File access level
`mysti.agents.autoSuggest`	`true`	Auto-suggest personas

Action	Windows/Linux	Mac
Open Mysti	`Ctrl+Shift+M`	`Cmd+Shift+M`
Open in New Tab	`Ctrl+Shift+N`	`Cmd+Shift+N`

Command	Description
`Mysti: Open Chat`	Open the chat sidebar
`Mysti: New Conversation`	Start fresh
`Mysti: Add to Context`	Add file/selection to context
`Mysti: Clear Context`	Clear all context
`Mysti: Open in New Tab`	Open chat as editor tab

Mysti collects anonymous usage data to improve the extension:

Feature usage patterns
Error rates
Provider preferences

No code, file paths, or personal data is ever collected.

Respects VSCode's telemetry setting. Disable via: Settings > Telemetry: Telemetry Level > off

Business Source License 1.1 (BSL 1.1)

Free for personal, educational, and non-profit use
Commercial use requires a separate license
Converts to MIT License on December 3, 2030

Contact baha@deepmyst.com for commercial licensing.

Install • Report Issue • GitHub

Mysti — Built by DeepMyst Inc
_{Made with Mysti/sub>}

Read the original article

Comments

By d4rkp4ttern 2025-12-2713:097 reply

A workflow I find useful is to have multiple CLI agents running in different Tmux panes and have one consult/delegate to another using my Tmux-CLI [1] tool + skill. Advantage of this is that the agents’ work is fully visible and I can intervene as needed.

[1] https://github.com/pchalasani/claude-code-tools?tab=readme-o...

By vidarh 2025-12-2718:181 reply

Have you considered using their command line options instead? At least Codex and Claude both support feeding in new prompts in an ongoing conversation via the command line, and can return text or stream JSON back.

By d4rkp4ttern 2025-12-2718:49

You mean so-called headless or non-interactive mode? Yes I’ve considered that but the advantage communication via Tmux panes is that all agent work is fully visible and you can intervene as needed.

My repo has other tools that leverage such headless agents; for example there’s a resume [1] functionality that provides alternatives to compaction (which is not great since it always loses valuable context details): The “smart-trim” feature uses a headless agent to find irrelevant long messages for truncation, and the “rollover” feature creates a new session and injects session lineage links, with a customizable extraction of context for the task to be continued.

[1] https://github.com/pchalasani/claude-code-tools?tab=readme-o...

By petesergeant 2025-12-2715:101 reply

I've had good success with a similar workflow, most recently using it to help me build out a captive-wifi debugger[0]. In short, it worked _pretty_ well, but it was quite time intensive. That said, I think removing the human from the loop would have been insanity on this: lots of situations where there were some very poor ideas suggested that the other LLMs went along with, and others where one LLM was the sole voice of reason against the other two.

I think my only real take-away from all of it was that Claude is probably the best at prototyping code, where Codex make a very strong (but pedantic) code-reviewer. Gemini was all over the place, sometimes inspired, sometimes idiotic.

0: https://github.com/pjlsergeant/captive-wifi-tool/tree/main

By bahaAbunojaim 2025-12-2715:21

This is exactly why I built Mysti because I used that flow very often and it worked well, I also added personas and skills so that it is easy to customize the agents behavior and if you have any ideas to make the behavior better then please don’t hesitate to share! Happy to jump on a call and discuss it as well

By bikeshaving 2025-12-2717:003 reply

I have a similar workflow except I haven’t put time into the tooling - Claude is adept at TMUX and it can almost even prompt and respond to ChatGPT except it always forgets to press Enter when it sends keys. Have your agents been able to communicate with each other with tmux send-keys?

By theturtletalks 2025-12-2717:37

I had the same issue. Subagents are nice but the LLM calling them can’t have a back and forth conversation. I tried tmux-cli and even other options like AgentAPI[0] but the same issue persists, the agent can’t have a back and forth with the tmux pane.

To people asking why would you want Claude to call Codex or Gemini, it’s because of orchestration. We have an architect skill we feed the first agent. That agent can call subagents or even use tmux and feed in the builder skill. The architect is harnessed to a CRUD application just keeping track of what features were built already so the builder is focused on building only.

0. https://github.com/coder/agentapi

By d4rkp4ttern 2025-12-2717:40

Yes this and other edge cases is why I made the Tmux-CLI wrapper. Yes they use send-keys with suitable delays etc

By zingar 2025-12-2717:211 reply

What are you asking/expecting Claude to do with tmux?

By bikeshaving 2025-12-2718:401 reply

I find that asking Claude to develop and Codex to review the uncommitted changes will typically result in high-value code, and eliminate all of Claude’s propensity to perpetually lie and cheat. Sometimes I also ideate with Claude and then ask Claude to get ChatGPT’s opinion on the matter. I started by copy-pasting responses but I found tmux to be a nice way to get rid of the middleman.

By joshstrange 2025-12-2814:321 reply

What does tmux add here? Or how does it allow you to do that? I’m sorry I’m just missing it I’m sure. I don’t use tmux a lot so I don’t know all its potential.

By aoeusnth1 2025-12-294:191 reply

It lets Claude directly type into Codex as if it were the user, or vice versa

By zingar 2025-12-2912:48

And you’re finding that it can’t do that without tmux?

By bahaAbunojaim 2025-12-2713:44

I will look it up indeed

By wild_egg 2025-12-281:231 reply

What does Tmux-CLI add on top of regular tmux?

Everything in the "What Claude Code Can Do With Tmux-CLI" section is already easily possible out of the box with vanilla tmux

By d4rkp4ttern 2025-12-282:36

You're right that vanilla tmux can do all of this, if a human were to use it. tmux-cli exists because LLMs frequently make mistakes with raw tmux: forgetting the Enter key, not adding delays between text and Enter (causing race conditions with fast CLI apps), or incorrect escaping.

It bakes in defaults that address these: Enter is sent automatically with a 1-second delay (configurable), pane targeting accepts simple numbers instead of session:window.pane, and there's built-in wait_idle to detect when a CLI is ready for input. Basically a wrapper that eliminates the common failure modes I kept hitting when having Claude Code interact with other terminal sessions.

By throwaway12345t 2025-12-2715:173 reply

This is cool, if Codex or Gemini CLI is supported it would be good to have a section in the readme indicating shortcomings etc (may have missed)

By tikimcfee 2025-12-2715:24

The idea works well with or without direct integration. You can have a cli agent read arbitrary state of any tmux session and have it drive work through it. I use it for everything from dev work to system debugging. It turns out a portable and callable binary with simple parameters is still easier to use for agents than protocols and skills: https://github.com/tikimcfee/gomuxai

By d4rkp4ttern 2025-12-2716:39

There’s no special support needed; it’s just a bash command that any CLI agent can use. For agents that have skills, the corresponding skill helps leverage more easily. I’ll add that to the README

By bahaAbunojaim 2025-12-2715:19

Claude code, Gemini and codex are all supported but need more testing so I would really value the feedback, bug reports and contributions as well :D

Contributions will be highly appreciated and credited

By sharifabdel 2025-12-2715:012 reply

What prompted you to build this?

By d4rkp4ttern 2025-12-2716:42

I have both Codex and Claude subs so I wanted one to be able to consult the other. Also it’s useful when you have a cli script that an agent is iterating on, so it can test it. Another use case is for a CLI agent to run a debugger like PDB in another pane, though I haven’t used it much.

By bahaAbunojaim 2025-12-2715:07

I used to get stuck sometimes with Claude and needing a different agent to take a look and the switch back and forth between those agents is a headache and also you won’t be able to port all the context so thought this might help solve real blockers for many devs on larger projects

By csar 2025-12-2717:201 reply

Getting feedback on a plan or implementation is valuable because you get a fresh set of eyes. Using multiple models may help though it always feels a bit silly to me (if nothing else you’re increasing non-determinism because you know have to understand 2 LLM’s quirks).

But the “playing house” approach of experts is somewhere between pointless and actively harmful. It was all the rage in June and I thought people abandoned that later in the summer.

If you want the model to eg review code instead of fixing things, or document code without suggesting improvements (for writing docs), that’s useful. But there’s. I need for all these personas.

By bahaAbunojaim 2025-12-2720:161 reply

The way it works is that each agent think independently, discuss the solution and each agent opinion then one will synthesize a solution.

By csar 2025-12-2722:081 reply

I understand. My point is that the personas are generally not a good idea and that there are much simpler and more predictable ways of getting better results.

By jacob019 2025-12-282:301 reply

I get where you're coming from, especially since role playing was so vital in early models in a way that is no longer necessary, or even harmful; however, when designing a complex system of interactions, there's really no way around it. And as humans we do this constantly, putting on a different hat for different jobs. When I'm wearing my developer hat, I have to reason about the role of each component in a system, and when I use an agent to serve in that role, by curating it's context and designating rules for how I want it to behave, I'm assigning it a persona. What's more, I may prime the context user and assistant messages, as examples of how I want it to respond. That context becomes the agent's personality--it's persona.

By bahaAbunojaim 2025-12-286:25

Spot on

By tombert 2025-12-284:221 reply

I so want to like these vibe coding agents, and sometimes I do, but it really does kind of suck the joy out of things.

What I was hoping would be that I could effectively farm out work to my metaphorical AI intern while I get to focus on fun and/or interesting work. Sometimes that is what happens and it makes me very happy when it does. A lot of the time, however, it generates code that is wrong, or incomplete (while claiming it is complete), and so I end up having to babysit the code, either by further prompting or just editing the code.

And then it makes a lot of software engineering become "type prompt, sit and wait a minute, look at the code, repeat", which means I'm decidedly not focusing the fun part of the project and instead I'm just larping as a manager who backseat codes.

A friend of mine said that he likes to do this backwards: he writes a lot of the code himself and then he uses Claude Code to debug and automate writing tedious stuff like unit tests, and I think that might make it a little less mind numbing.

Also, very tangential, and maybe my prompting game isn't completely on point here, but Codex seems decidedly bad at concurrent code [1]. I was working on some lock-free data store stuff, and Codex really wanted to add a bunch of lock files that were wholly unnecessary. Oh, and it kept trying to add mutexes into Rust, no matter how many times I tell it I don't want locks and it should use one-shot channels instead. To be fair, when I went and fixed the functions myself in a few spots and then told it to use that as an example, it did get a little better.

[1] I think this particular case is because it's trained on example code from Github and most code involving concurrency uses locks (incorrectly or at least sub-optimally). I guess this particular problem may be more of the fault of American universities teaching concurrent programming incorrectly at the undergrad level.

By bahaAbunojaim 2025-12-286:331 reply

I find it useful to let one agent come up with a plan after a review and another agent implementing the plan. For example, Gemini reviewing the code, codex writing a plan and then Claude code implementing it

By mycall 2025-12-2815:061 reply

What about the reverse, after Claude code implements it, let Gemini/Codex do a code review for bugs and architecture revisions? I found it is important to prompt to only make absolutely minimal changes to the working code, or unwanted code clobbering will happen.

By bahaAbunojaim 2025-12-297:35

That works great too. Will be adding the ability to tag another agent in a near release

Hacker News