CODEX SKILL · LOCAL WORKSPACE

One person.
A whole AI team.

KINE TEAM brings your Codex agents into one familiar workspace. Keep a conversation going, bring the right teammates into a group, and follow the work in one place.

COMING SOON In development for Codex on macOS

Different roles.
One place to work together.

A conversation, with a team behind it.

DEVELOPMENT PREVIEW
KINE TEAM's development workspace: a conversation list on the left, group messages, delegation progress, and a composer on the right.
Group conversations, work progress, and replies in one view. Test data · UI preview

Screenshots use isolated test conversations and simulated work states to show the interface, not live model execution.

01 / CONTINUITY

Pick up where you left off.

Return to the same teammates and conversations across Codex tasks. Keep a lasting one-to-one chat with an agent, without starting over each time.

02 / COLLABORATION

Bring the right people together.

Create a group around a goal. A team lead coordinates research, planning, development, and review, with progress and replies kept in the conversation.

03 / PROJECT CONTEXT

Keep the work in its place.

Choose the project, attach files, and use Codex Skills in context. Review permissions and approvals as part of the conversation.

A TEAM YOU CAN SHAPE

Familiar teammates.
Memory you can manage.

Give each agent a role and a way of working. Inspect, edit, or remove remembered preferences and decisions, and choose whether an agent saves memory automatically.

Project-specific memory stays scoped to its project. Conversations and memory are separate: not every message becomes a lasting note.

An agent's memory settings showing an automatic-memory toggle, project scope selector, and editable memory field.
Agent settings and memory controls. Test data · UI preview
COMING SOON

A workspace for your next team.

KINE TEAM is being prepared for public release as a Codex Skill with a local runtime, not a standalone Mac app. There is no public download yet.

The workspace runs locally; AI execution still uses Codex and any explicitly authorized providers. Features and interface may change before release.