@everyone New proposal: unlimited Traycer at just $10/month Bring your own inference through your Claude Code or Codex plans. Traycer’s UX changes by replacing the native chats with CC or Codex TUIs embedded right in the sidebar. This also means you can run implementation agents right inside the sidebar instead of being restricted to only planning in Traycer. For users who prefer Traycer chat interface can continue to use it on the higher plans. Let me know what you guys think in thread replies. React with ➕ or ➖ based on whether you are for or against this change

you guys need to make some sort of CLIf or traycer. otherwise, this sounds great

Would also (for work purposes) love to be able to bring in my GitHub copilot enterprise- would that be possible as currently envisioned? I don’t have access to the cli executable, just the vs code chat

We can support GitHub/Copilot CLI

But including a vs code to vs code handoff to copilot? The environment permits the vs code extension but not the new copilot cli

The handoffs will work as they do today. You can create custom CLIs in Traycer if something is not supported out of the box https://docs.traycer.ai/integrations/custom-cli-agents

Apologies!! In my personal capacity I am all for that

@flat-river-95835 with traycer being ran in the cli along something like CC would that mean the "YOLO" mode can essentially be 100% automated with claude responding to follow up questions automatically and allowing traycer and the agent to handle most of the workflow 100% in the CLI

Yes, you can do 100% of the Traycer workflow directly in the CLI (Claude Code/Codex) by simply configuring our upcoming MCP/Skill And you can pair it with a web browser to look at specs and tickets. You don't need the extension

This will be huge!!

So 10 a month to use an empty ai less platform with the goal ultimately being for us to populate it with the other ais we also pay 20-40 / mo for separately..... So the path would be : Use our existing ais to infer and tell itself about the work its gonna have to do, then have the same ai check its work and validate itself... What could go wrong there?

@purple-sugar-95000, you can mix and match multiple CLIs and models. E.g., CC’s Opus to plan, CC’s Sonnet to write code, Codex’s GPT-5.4 to verify the changes If you have a Cursor subscription, you can use the cursor-agent CLI with different models at each step. This is even more powerful than what you can achieve in Traycer today. You have the freedom to choose a harness, not just models. And can leverage inference from any subscription.