Compare

my-ai-team is a managed delivery loop that runs on top of the agent CLI you already use, so it differs from driving a coding agent yourself.

The comparison is about the working style of each kind of tool, not the features of a particular product or version.

Capability	Hand-run CLI agent (Claude Code, Codex CLI)	IDE agent	my-ai-team
Runs unattended on a backlog	You start each task yourself in an interactive session.	You start each task yourself from inside the editor.	Agents select the next ready ticket and begin without you.
Issue handoff via labels	You paste the task into the session by hand.	You describe the task in the editor’s assistant.	Marking an issue ready hands it off; claiming it swaps the label for a mode lock.
Role separation across modes	One session plays every role.	One assistant acts across the whole workflow.	Planning, implementation, and review run as separate modes and backend slots.
Remote control over Telegram	The terminal is the only control surface.	The editor is the only control surface.	An optional relay puts every notification and reply in Telegram.
Self-merging pull requests	You open the pull request and merge it yourself.	You merge from the editor’s source-control view.	A separate reviewer gates the work and the merge happens without you.
Runs on top of your existing backend	The CLI is the backend you are driving.	The tool brings its own model access.	It runs on top of claude, codex, copilot, pi, opencode, freebuff, or cmd (Command Code).

my-ai-team is not an editor, not a model, and not a replacement for the backend CLI. It is the delivery loop around a backend you already run, with your own account and token.