ACP UIvsAgent Maestro
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
ACP UI
Jun Han · Agent harness
OSS
Panel
6.31 spec wins
- Reliability
- 5.8
- Usefulness
- 5.7
- Cost
- 8.3
- Longevity
- 5.2
“There is an Android build, so you can now approve a file write from the bus and regret it before your stop.”
Agent Maestro
Joouis · Agent harness
OSSMCP
Panel
6.33 spec wins
- Reliability
- 5.8
- Usefulness
- 6.0
- Cost
- 8.3
- Longevity
- 5.0
“Its hosted web-search tool is executed through Exa, so your editor now subcontracts curiosity.”
Spec by spec
| Spec | ACP UI | Agent Maestro |
|---|---|---|
| Architecture | ||
| Category | Agent harness | Agent harness |
| Runs | local | local |
| Platforms | macos, linux, windows, web | macos, linux, windows |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | No | Yes |
| MCP server | No | No |
| Capabilities | ||
| Runs terminal commands | YesAs the ACP client, ACP UI executes the terminal commands the agent requests and shows their output; the desktop build only, since the browser cannot spawn processes. | YesAgent Maestro is a control plane; command execution and file edits happen inside the Roo Code, Cline or CLI agent it drives. |
| Multi-file edits | YesFile reads and writes are performed by ACP UI on the agent's behalf against the host workspace, and only on desktop. | Yes |
| Git operations | No | No |
| Browser control | No | No |
| Sandboxed execution | No | No |
| Multi-agent orchestration | No | YesUp to twenty concurrent Roo Code tasks can run at once. |
| Headless / CI mode | No | YesTasks are created and managed entirely through REST APIs, which the README calls headless AI agent control. |
| Models | ||
| Backbone | Claude Code, Codex, GitHub Copilot, Gemini CLI, Qwen Code, OpenCode, OpenClaw, Kiro CLI, any ACP-compatible agent | Anthropic, OpenAI, Gemini, GitHub Copilot |
| Bring your own model | Yes | Yes |
| Local models | No | No |
| Cost | ||
| Pricing model | byok | byok |
| Starts at | $0/mo | $0/mo |
| Free tier | Yes | Yes |
| Bring your own key | Yes | Yes |
| Openness | ||
| Open source | Yes | Yes |
| License | MIT | MIT |
| GitHub stars | 491 | 206 |
Which one would each critic pick
| Critic | ACP UI | Agent Maestro | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 6.5 | 6.8 | Agent Maestro — Pick this if you already run Roo Code or Cline in VS Code and want a script to start their tasks; pick a plain terminal agent if you would rather the editor were gone. |
| El Crítico | 6.0 | 6.0 | no preference |
| El Profesor | 6.5 | 6.8 | Agent Maestro — Server-Sent Events for live monitoring and an OpenAPI document at /openapi.json mean the control surface is specified rather than described. |
| La Inversora | 5.5 | 5.5 | no preference |
| La Jefa | 6.0 | 5.0 | ACP UI — Nothing per seat across sixty desks and nothing to govern: no console, no single sign-on, no audit record, and nothing that runs in a pipeline. |
| El Hacker | 7.0 | 7.8 | Agent Maestro — MIT, and it flips the arrow: Claude Code, Codex or Gemini CLI point at my editor instead of a provider, because it answers /messages and /chat/completions itself. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.