Team Mode skill
Team Mode splits a Codex task between an explorer, an executor and a reviewer while the main agent keeps final acceptance
- 235 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The skill gives subagents the right to edit project files and the roles in `.codex/agents` change Codex settings.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/team-modeDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill gives subagents the right to edit project files and the roles in `.codex/agents` change Codex settings.
This is third-party code. Review the repository files before installing.
What it does
Team Mode is a Codex skill that helps the main agent delegate work to subagents with clear roles. The main agent breaks the task down, makes the open calls and accepts the result. Simple tasks can run without a single subagent. There are four roles. Explorer finds relevant files, entry points and relationships in a large codebase. Executor implements where the goal and file ownership are clear. Reviewer independently checks a stable result with fresh context and can save a Markdown report. ExpertAdvisor takes complex decisions, modeling and hard cases, and its model is chosen per run. Inside the skill sit the Explore and Simplify references: the first finds relevant code, the second checks change complexity before commit. Dispatch rules forbid subagents from launching other agents and parallel tasks from sharing write ownership. Roles are configured with `.toml` files in `agents`.
Who it is for. Codex users who work on complex tasks in large repositories and want to split exploration, implementation and independent review.
Good fit when
- You need to find an implementation in a large repository fast without loading the main context.
- The work splits into independent parts with clear files that Executors do in parallel.
- A stable result needs independent review from a clean slate before acceptance.
- Before committing you want to check whether a change got too complex: the Simplify reference fits.
Not a fit when
- The task is simple and fits one context: delegation adds cost with no gain.
- You work outside Codex: the skill and roles target its subagent format.
- You cannot change global Codex settings: the optional `default.toml` overrides the built-in `default` agent.
Example request
Use `$team-mode` for this task. Delegate to subagents only the parts with clear benefit and keep acceptance for yourself.Limitations
The skill and role configs install separately: `npx skills add oil-oil/codex-team-mode` for the skill, while `Explorer.toml`, `Executor.toml`, `Reviewer.toml` and `ExpertAdvisor.toml` are copied to `~/.codex/agents/` or to the project `.codex/agents/`. Model choices sit in the `.toml` files, and a pinned model and reasoning level there override values passed at dispatch, so ExpertAdvisor is left unpinned. The optional `default.toml` affects all Codex tasks and is not part of the default install. The cost estimate in `usage_by_model.py` is approximate and is not a bill. The documentation is written in Chinese.
How to disable. Remove the skill with the `npx skills` manager and the role files from `~/.codex/agents/`. The optional `default.toml` can be moved to `~/.codex/agents-disabled/`.
FAQ
How do I install Team Mode?
Run `npx skills add oil-oil/codex-team-mode`, then copy the role files from `agents/` into `~/.codex/agents/`.
Which roles does the skill have?
Explorer, Executor, Reviewer and ExpertAdvisor. The first three get fixed settings, ExpertAdvisor picks its model per run.
How do I invoke the skill?
With a phrase such as `Use $team-mode for the task`. The main agent decides which subagents to start.
Can subagents launch nested subagents?
No. Dispatch rules forbid subagents from launching others and the main agent accepts the result.
How do I see usage per model?
With `python3 skills/team-mode/scripts/usage_by_model.py --days 7 --by-agent --json`: it counts from local logs and the numbers are approximate.
Discussion
No questions or reviews yet.
Related
SkillsA skill that hands work to a team of a CTO, five PMs and fifty engineers: roles, permissions, a task log and resource leases.
Viden is a terminal cockpit for a coding agent: chat with a model, approve edits and supervise Codex and Claude in one window.
A Codex skill collection: phased planning, parallel subagents, documentation access, a browser, 51 hooks and ready agent roles.
A meta-skill that splits a task into stages, picks a profile and runs multi-agent work in Codex and Claude Code.
A Codex skill that splits work across three models: the main thread plans, Terra handles the screen, Luna runs the tests.
A Codex skill for an autonomous loop: Claude Code computes, Codex reviews hourly, critiques and sends the next prompt.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Team Mode (this card) | no | 235 | |
| Codex Astra + Luna Orchestrator | no | 1,748 | Codex Astra Orchestrator builds orchestration around its own subagents, while Team Mode defines four roles with Explore and Simplify references. |
| Awesome Codex Subagents | no | 6,473 | awesome-codex-subagents gives a collection of ready subagents, while Team Mode gives a delegation scheme with acceptance by the main agent. |
README in short
The README covers roles, dispatch boundaries, installing the skill and roles, and model and usage diagnostics from local logs. MIT license, Chinese and English versions.