Debate
"I want two positions to confront each other."
Two agents answer in turn, then a summary sets out agreements and disagreements.
01 / PALABRE
Orchestrate your AI agents. Keep the decision.
PALABRE lets the AI agents already installed on your machine work together. It helps you make better decisions before acting, and orchestrates the exchanges between your agents.
npm install -g palabre02 / RELAY
Orchestrate the work between your agents
With Relay mode, one agent sends a message directly to another agent, in another conversation that already has your project in context. It replies, and the work is shared between them. You are no longer the go-between, but the decision-maker.
- Agent A
- PALABRE
- Agent B
- reply
- you decide
Fictional example: lead agent and teammate
You hand a feature to Claude Code, your lead agent. It splits the work and hands it, step by step, to Codex, its teammate, which already knows the code. Codex prepares a PR; Claude reviews it and asks for a fix; Codex fixes it. At every important decision, the lead agent stops and asks you. You copy nothing: you approve.
Guarantees
- The right context. The message lands in an existing conversation, with everything it already knows.
- You stay in control of permissions. The message asks; it grants nothing more. A closed conversation replies in read-only mode; an open Codex conversation works with the permissions you gave it.
- Everything is logged. One reply per message, no automatic resend, and a
.relay.mdexport.
Relay connects Claude Code and Codex both ways, in the CLI as well as on desktop. Toward a conversation that is already open, only Codex can receive the task for now, on Windows. To chain several steps within a fixed scope, Relay offers the Mission option.
03 / DEBATE
Confront opinions before you decide
With Debate mode, two agents answer the same question in turn. Each one reads the other's answer, challenges or completes it. At the end, a summary sets out agreements, disagreements and proposed actions.
- Topic
- Agent A
- Agent B
- summary
- you decide
Fictional example
You are unsure whether to split a module that has grown too large. Codex, as implementer, proposes a split into three parts; Claude Code, as reviewer, spots two hidden dependencies. Codex adjusts. The summary keeps the split, flags the remaining risk and suggests a first test to write. You decide, with the arguments in hand.
- Two perspectives, not an echo. Roles steer each agent: one proposes, the other looks for what could break.
- A useful summary. Consensus, disagreements and uncertainties, proposed actions, conclusion.
- The right pace. From 1 to 20 responses. The debate stops on its own as soon as an explicit agreement is reached.
04 / MODES
Which mode for what?
Palabre offers four modes: Debate, Relay, Ask and Chat.
Relay
"I want my agents to cooperate and hand off to each other, without any copy-paste from me."
One agent sends a message directly to another agent (Claude Code or Codex), which replies. You are no longer the go-between, but the decision-maker.
Ask
"I want several independent opinions."
1 to 4 agents answer without seeing each other's answers, then a summary compares them.
Chat
"I'm chatting with an agent, and sometimes need a second opinion to double-check an answer."
Ask another agent with /consult, without leaving the conversation.
In doubt: Debate for a decision with stakes, Relay to orchestrate work across several agents, Ask for several opinions, Chat to talk with one agent.
05 / AGENTS
Your agents, the way you already use them
Claude Code, Codex, OpenCode, Mistral Vibe, Antigravity and Ollama. PALABRE runs their CLIs with your accounts, subscriptions and models: if an agent works in your terminal, it works with PALABRE. Your Claude Code and Codex conversations, in the CLI as well as in the desktop apps, can receive a Relay. On Windows, Codex desktop even receives the message in the open conversation.
06 / PRIVACY
Your data only goes through your agents.
Your data only goes to the agents you choose: it never passes through a PALABRE server. With Ollama and a local model, nothing leaves your machine.
07 / GET STARTED
Get started
Check your agents.
palabre doctor checks the configuration and lists the agents found on your machine.
Start a debate.
palabre codex-claude "Should we split this module?": two agents, one question. Without arguments, palabre opens the interactive interface.
Attach files.
--files src/auth.ts README.md sends exactly the files you choose.
Attach folders.
--context src docs scans the folders and leaves out files that are too large or binary.
Give a role.
In Debate, Ask or Chat, you can give each agent a role that steers its instructions: implementer, reviewer, critic, architect or scout.
Keep a record.
Each session leaves a Markdown file (.debate.md, .relay.md, .ask.md, .chat.md) to reread, share or hand to another agent.
08 / WHERE TO USE IT
Where you already work
Terminal (CLI)
the palabre CLI, with its interactive interface or commands to use in your scripts.
CLI agents
with the PALABRE skill, Claude Code, Codex and other CLI agents know when to consult another agent, and with which mode.
Desktop apps
from Claude Code desktop or Codex desktop, your agent runs the PALABRE CLI through the skill, and can hand off to another conversation.
VS Code
a sidebar to start a Debate, Chat or Ask and follow the session live.
One project, several agents, your decisions.
Install PALABRE, pick your mode, orchestrate your agents' work and keep the decision.