Team Avatar plugin
Orchestrator for Claude Code, Codex and Cursor: nine Avatar characters, 58 engineering skills and one routing mode.
- 113 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The mode launches sub-agents that edit repository files and run reviews.
- The Cursor edition installs through `curl | bash`.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/team-avatarDetects 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 mode launches sub-agents that edit repository files and run reviews.
- The Cursor edition installs through `curl | bash`.
This is third-party code. Review the repository files before installing.
What it does
A plugin with a single `avatar-mode` that picks a workflow and hands tasks to characters. Katara fixes bugs, Toph reviews the diff, Zuko owns performance and incidents, Iroh writes the docs. There are 58 skills in `SKILL.md` format. The base is the pstack skill set with characters on top. Each character runs on the model named in a model sheet that `avatar-setup` writes. A separate Cursor edition exists, the original oh-my-cursor. Other harnesses (OpenCode, Hermes Agent, Gemini CLI and more) install the skills through `npx skills add`.
Who it is for. Developers who want the habits of reproducing the bug, writing a regression test and reviewing the diff baked into a single mode on top of Claude Code or Codex.
Good fit when
- A team gets three bug reports and wants each to go through a test and review.
- A developer asks to explain an unfamiliar repository's structure without changes.
- A task must be split among several sub-agents with different models.
- A Cursor project moves to the plugin from the older oh-my-cursor edition.
Not a fit when
- You expect higher accuracy: on SWE-bench Verified Mini the plugin solved about as many tasks as pstack and no plugin, at higher cost.
- The environment forbids running several sub-agents.
- Running scripts through `curl | bash` for Cursor is unacceptable.
Example request
/team-avatar:avatar-mode Explain this repository's structure and how to run its tests. Cite the files. Change nothing.Limitations
The authors measured quality on SWE-bench Verified Mini; differences between Team Avatar, pstack and no plugin are not statistically significant, and per-task cost is higher. In Codex, sub-agents run only with `multi_agent = true` in `~/.codex/config.toml`, and default models are named for Claude, so run `$avatar-setup` once. The Cursor edition installs with `install.sh` through `curl | bash`: read it first.
How to disable. Run `claude plugin uninstall team-avatar@team-avatar` or remove the plugin in Codex. Remove the model sheet and the `multi_agent` setting manually.
FAQ
What is `avatar-mode`?
A single entry point: the mode picks the workflow and brings in the right characters. In Claude Code it is `/team-avatar:avatar-mode`, in Codex `$avatar-mode`.
How do I install without plugins?
Use `npx skills add tmcfarlane/team-avatar/plugins/team-avatar` or copy folders from `plugins/team-avatar/skills/` into the harness skills directory.
Does it improve results?
Not by the authors' measurements on SWE-bench Verified Mini. The gain is in habits: reproduce the bug, write a test, review the diff.
Where do the skills come from?
The base is Lauren Tan's pstack via Michael Denyer's Claude Code and Codex port; sources are credited in `NOTICE`.
What does Codex need?
Set `multi_agent = true` under `[features]` in `~/.codex/config.toml` and run `$avatar-setup` once: default models are named for Claude.
Discussion
No questions or reviews yet.
Related
PluginsA collection of 14 portable skills for long agent runs: work planning, live app QA, search across past sessions and a phone call
A process kit for agents: a skill router, project memory in a DEV folder and result verification before saying done.
One agent launches others in the background: Claude Code, Codex, Copilot CLI, Grok Build and Pi, with a model chosen per stage.
A desktop app where Claude Code, Codex, Gemini and other agents work as a team of specialists in a shared project
A plugin of seven skills for Claude Code and Codex: prompts, subagents, skill authoring, editing, web search and a browser.
Argus runs long engineering and research tasks with four roles, where a reviewer reads results but edits nothing.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Team Avatar (this card) | no | 113 | |
| pstack for Claude Code and Codex | no | 4,263 | a pstack port for Claude Code and Codex; Team Avatar adds characters, a routing mode and a model sheet |
| Superpowers | no | 316,679 | a general development workflow set; here roles are assigned to nine characters with separate models |
README in short
The authors state plainly that they added characters on top of pstack with no score gain. Window captures from Claude Desktop and the Codex app serve as demos.