mcpField
T
team

team

github.com/akillnessAgent workflows52109 installsUpdated today
Repository

About team

team is a skill for coordinated, runtime-backed work using omx and tmux. It guides teams that need shared runtime state or task lists through preflight checks, startup evidence, monitoring, and orderly shutdown.

Key features

  1. 01Checks for omx and tmux before starting
  2. 02Starts workers through omx team commands
  3. 03Captures team, pane, and state evidence
  4. 04Monitors tasks until they reach a terminal state
  5. 05Keeps runtime-owned state under its designated path

Use cases

  1. 01Coordinating multiple workers on a task
  2. 02Running team mode with a shared task list
  3. 03Checking team status and shutting down after work completes

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/akillness/jeo-skills --skill team

After installing, type/teamin Claude Code to confirm it loaded.

Bundled filesevals/evals.jsonSKILL.toon

Details

From the project README

Team

Use this exact-name skill for $team so Codex-style loaders can activate the coordinated-worker contract directly instead of relying on the broader omx router description.

When to use this skill

  • The user explicitly invokes $team.
  • The user asks for team mode, tmux workers, or coordinated multi-agent work.
  • Workers need shared runtime state or a shared task list.

Instructions

Follow the preflight, invocation, lifecycle, and route-out rules below.

Preflight

Before starting a real team runtime, check:

command -v omx
command -v tmux

If either is missing, stop and report the missing runtime requirement. Do not simulate omx team with unrelated in-process fan-out when the user asked for the runtime-backed team workflow.

Invocation

Use the shell runtime when available:

omx team 3:executor "task"
omx team ralph "task"
omx team status <team-name>
omx team shutdown <team-name>

Lifecycle Rules

  1. Start the team and capture concrete evidence: team name, tmux panes, and state path.
  2. Monitor with omx team status <team-name>.
  3. Wait until tasks are terminal before shutdown unless the user explicitly aborts.
  4. Keep .omx/state/team/<team>/ owned by the runtime; do not delete it manually.

Route-Outs

  • Use ultrawork for independent lanes with little coordination.
  • Use ultraqa for QA/review fan-out.
  • Use omx for installation, doctor checks, and cross-runtime mapping guidance.

Examples

  • $team 3:executor "fix all TypeScript errors"
  • omx team ralph "ship the auth cleanup with verification"

Best practices

  • Capture startup evidence before assuming the team is running.
  • Do not shut down while workers are still writing task updates.
  • Keep runtime state under .omx/state/team/<team>/.

References

  • omx for runtime setup and doctor checks.
  • ultrawork for independent burst lanes.

Comments

Markdown supported