Supported agents
An agent card runs a real AI coding agent — the same program you’d run in a terminal, with your own account. NeuroSquad supports ten of them, plus ordinary terminals.
| Agent | Arrows give abilities | Picks up where it left off | Dangerous mode |
|---|---|---|---|
| Claude Code | Yes | Yes | Yes |
| Codex CLI | Yes | No | Yes |
| Qwen Code | Yes | Yes | Yes |
| Cursor CLI | No | No | Yes |
| OpenCode | No | No | Yes |
| Hermes Agent | No | No | Yes |
| Kilo Code | No | No | No |
| Pi | No | Yes | No |
| omp | No | No | Yes |
| Crush | No | No | Yes |
The agent can use connected browsers, terminals, notes and other agents, and work in canvas mode. See Arrows.
After you restart NeuroSquad, the agent continues the same conversation. Agents without this start a fresh conversation each time.
The agent can skip asking for permission. See Dangerous mode.
Claude Code has the deepest support: it tells NeuroSquad exactly when it is waiting for your answer, and it has the context meter, Compact and the journal. Token usage of Claude Code, Codex CLI, OpenCode, Pi and omp shows up in Usage & costs.
Claude Code, Codex CLI, OpenCode and Qwen Code can also run on models from OpenRouter instead of your own login.
Terminals
You can also add plain terminals — your system shell, Windows PowerShell, PowerShell 7, Command Prompt or Bash (Git Bash). See Terminal card.
Not sure which agents you have? Settings → Harnesses lists what NeuroSquad found on your computer, and Settings → Setup can install one for you.