Voff CLI
A terminal app that builds a team of Codex, Claude and Ollama agents and runs tasks in parallel under a lead agent
- 92 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Agents read project files and propose edits and local commands, which apply only after your approval.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/voffDetects 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.
- Agents read project files and propose edits and local commands, which apply only after your approval.
This is third-party code. Review the repository files before installing.
What it does
Voff is a terminal interface for team work of coding agents written in Rust. Running `voff` in a project directory creates a team of agents for that project. You send a task to the lead, it splits the work between agents, they carry out independent parts in parallel and the lead then checks each result against acceptance criteria, asks for fixes and integrates the verified work. One team can mix providers and models from Codex, Claude and Ollama and build nested teams. Agents change nothing on their own: a file edit (`file.patch`) and a local command (`process.test`) arrive as proposals and pass an approval before reaching the workspace. The other tools are `file.read`, `search.text`, `task.delegate`, `review.request`, `task.complete` and `task.blocked`. The interface shows a team map and tasks live and counts token use and cost estimates per agent. The team is saved in `.voff/config.toml`, and `voff --setup` reruns setup while keeping the task history.
Who it is for. Developers who live in the terminal and want to watch several agents split a task while approving each edit and command separately.
Good fit when
- A task splits into independent parts that need to run in parallel with different agents
- One team needs models from different providers, for example Claude for planning and a local Ollama model for routine work
- Every file change and every local command must be approved by hand before it applies
- You want to see token use and rough cost per agent
- An agent result has to pass a lead review against acceptance criteria with a request for fixes
Not a fit when
- The task is small and fits one conversation with one agent: a team of several agents only adds cost
- You need work without a terminal, for example in a web interface: Voff is terminal-only
- Agents need freedom without approval: in Voff every file edit and local command goes through confirmation
Example request
Add a CSV export command to the project: let one agent write the code, another the tests and the lead check the result against acceptance criteria.Limitations
Rust and Cargo or an Arch Linux package are required: install with `cargo install voff` or `pacman -S voff`. For agents to work the chosen providers must be available: Codex, Claude or Ollama, and their access is configured separately. The team and state live in `.voff/` inside the project, so add the directory to git ignore if you do not want it in the repository. Parallel work by several agents raises token use, and the per-agent cost estimate stays approximate.
How to disable. Remove the binary with `cargo uninstall voff` and the `.voff/` directory from the project. Accepted edits stay in the working copy and roll back through git.
FAQ
How do I install Voff?
Run `cargo install voff` if Rust is installed, or `pacman -S voff` on Arch Linux. Then start `voff` in the project directory.
Which providers are supported?
Codex, Claude and Ollama, and their models can be mixed in one team. Nested teams let work be delegated deeper.
Can agents change files without approval?
No. A file edit and a local command arrive as proposals and apply after approval. Reading files and searching text have the `file.read` and `search.text` tools.
How do I change the team?
Edit `.voff/config.toml`, an example sits in `config/example.toml`. `voff --setup` reruns setup and keeps the project task history.
Which keys control the interface?
1 and 2 or `Tab` switch tabs, `Enter` sends a task to the lead, `a` adds an agent, `o` opens the selected agent tab and `?` shows help.
Discussion
No questions or reviews yet.
Related
CLI toolsViden is a terminal cockpit for a coding agent: chat with a model, approve edits and supervise Codex and Claude in one window.
A Rust CLI where a meta-agent hands work to Claude Code and Codex, each on its own branch and tmux session, in one view.
A harness around a coding agent: supervisor, reviewer, adversary and local log distiller, each on its own model
multi-agents runs a team of coding agents on a task and measures whether it beat a single agent and what it cost.
A planner for Claude Code, Codex and OpenCode: one goal becomes a task plan and independent tasks run in parallel
A visual workspace for teams of Claude Code, Codex, OpenCode, Grok and Pi agents: files, Git, terminals and tasks in one window
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Voff (this card) | no | 92 | |
| Claude Squad | no | 8,790 | runs several agent sessions in separate worktrees; Voff builds a team with a lead, acceptance checks and edit approval |
| Agent Deck | no | 1,252 | a session dashboard; Voff splits a task between agents and merges the result through a lead |
README in short
The README shows a quick start, features (mixed providers, nested teams, cost tracking), the flow of a lead with agents and approval, key bindings, a tools table and configuration through `.voff/config.toml`.