Keep Kiro.
Give it your machines.
AI Commander is not an alternative to Kiro, Claude Code, Codex, ChatGPT, Cursor, GitHub Copilot, Sourcegraph Cody, Grok, Gemini, Windsurf, Continue, Cline, Roo Code, Zed, OpenCode, Qwen Code, goose, Amp, Kilo Code, OpenHands, Amazon Q Developer, Augment Code, Devin, or JetBrains Junie. It is the remote-machine harness those tools plug into over MCP: outbound-only shell, detached jobs that outlive the chat, and file transfer on hardware you already own.
- Install the host agent on the machine you want your AI to control. It shows a private machine code.
- Connect your AI: paste this prompt, then replace
AIC-…with your code —Help me connect this AI client to AI Commander using https://aicommander.dev/start.md. Then check the status of my machine: AIC-…— or add the connector URLhttps://aicommander.dev/mcpand sign in with an email code (per-client steps).
Sibling harness pages: Junie · Cursor · GitHub Copilot · Cody · OpenHands · Amazon Q · Augment · Devin · Aider · Cline · Roo Code · Zed · OpenCode · Qwen Code · goose · Amp · Kilo Code · Continue · Windsurf · Claude Code · Grok · Codex · ChatGPT · Gemini · Cua Spaces · Warp.
Connect Kiro (remote MCP)
Kiro, the agentic IDE and CLI
(Kiro MCP configuration docs),
reads MCP servers from an mcp.json file. Per Kiro’s docs, remote servers are declared with a url
(plus optional headers), and Kiro handles the browser-based OAuth flow for servers that use Dynamic Client Registration.
Point it at AI Commander’s hosted MCP endpoint:
{
"mcpServers": {
"aicommander": {
"url": "https://aicommander.dev/mcp"
}
}
}
Use ~/.kiro/settings/mcp.json for user scope (all workspaces) or .kiro/settings/mcp.json in the workspace for project scope —
if both exist, Kiro merges them and the workspace entry wins. In the IDE, open the command palette and run
Kiro: Open user MCP config (JSON), or use the Open MCP Config icon in the Kiro panel; saving the file reconnects servers.
Do not commit secrets: if you add an Authorization header with an API key, reference an environment variable
(for example "Bearer ${AIC_API_KEY}") and approve it in Kiro’s Mcp Approved Env Vars setting.
Prefer a local stdio bridge?
{
"mcpServers": {
"aicommander": {
"command": "npx",
"args": ["-y", "@aicommander/mcp"],
"env": { "AICOMMANDER_SERVER": "https://aicommander.dev" }
}
}
}
Keep tools behind Kiro’s approval prompt at first; only add AI Commander tools to autoApprove once you trust the workflow.
You can also hide tools you do not want an agent to call with disabledTools.
Full client matrix (Kiro, Warp, Junie, Augment, Devin, Claude Code, Codex, ChatGPT, Cursor, Copilot, Cody, Grok, Gemini, OpenCode, Qwen Code, goose, Amp, Kilo Code, Continue, Windsurf, Cline, Roo Code, Zed, OpenHands, Amazon Q, REST): Connect your AI client.
Over the remote URL, first connect may open a one-time browser sign-in (OAuth Dynamic Client Registration).
The npx bridge does not sign in through a browser: without AICOMMANDER_TOKEN in its env it connects anonymously with a code-only first hour; set it to your API key to bind it to your account.
On the remote URL, append ?anonymous=1 to skip the sign-in, for a code-only first hour.
For optional account features (saved aliases), pass a Bearer API key in headers.
Menu names and options can differ between Kiro versions — check the linked Kiro docs for your build.
What you get that the chat alone cannot
- Remote shell on Linux, macOS, and Windows — by session code or saved alias.
- Detached jobs for builds, test suites, and training on the box that has the repo.
- Fleet list + GPU facts so Kiro picks the right machine before it burns an hour.
- File courier (Pro) for checkpoints and diffs up to 100 MiB.
Not a replacement — a multiplier
| You already use | AI Commander adds |
|---|---|
| Kiro | mcpServers.aicommander.url → https://aicommander.dev/mcp in ~/.kiro/settings/mcp.json |
| Warp | Settings → Agents → MCP servers → + Add → https://aicommander.dev/mcp |
| JetBrains Junie | mcpServers.aicommander.url in ~/.junie/mcp/mcp.json |
| Augment Code | auggie mcp add aicommander --transport http --url https://aicommander.dev/mcp |
| Devin | Settings → MCP → custom HTTP → https://aicommander.dev/mcp |
| Claude Code | claude mcp add --transport http … → remote exec + jobs |
| Codex / ChatGPT | Same MCP endpoint or REST /api/v1/exec |
| Cursor / Continue / Windsurf | npx @aicommander/mcp stdio bridge |
How Kiro handles MCP servers
- MCP support has to be switched on in Kiro's settings first.
- Kiro reads
~/.kiro/settings/mcp.json(user) and.kiro/settings/mcp.json(workspace) and merges them. The workspace wins on a name clash, and anmcpServersblock in an agent config overrides both. autoApprovelists tools that run without a prompt ("*"means all of them), anddisabledToolshides tools.${VAR}in a config is only expanded for variables you have approved in the "Mcp Approved Env Vars" setting.- Saving
mcp.jsonreloads it right away and only restarts the servers that changed. - Remote OAuth uses dynamic client registration with the scopes
openid email profile offline_access. If a server rejects those scopes, set"oauthScopes": []. The CLI can re-authenticate with/mcp authand sign out with/mcp logout.
Checked against Kiro docs: MCP configuration in November 2026. Client behaviour changes between releases; the official docs win if they disagree.
A workflow that suits Kiro
Kiro is spec-driven: it writes requirements and a task list before coding. AI Commander lets the tasks that need real hardware run on it.
- Add
aicommanderto the workspace.kiro/settings/mcp.jsonwithautoApproveforlist_machines,session_statusandremote_job_status. - In the spec, write the verification task plainly, for example "run the load test on the staging box as a background job and record p95 latency".
- When Kiro reaches that task it calls
remote_job_start, then readsremote_job_logsand writes the numbers back into the task notes.
Limits and gotchas in Kiro
- An API key referenced as
${AICOMMANDER_KEY}stays empty until that variable is added to Mcp Approved Env Vars. - Avoid
"autoApprove": ["*"]for AI Commander. It would also approveremote_execon every machine. - If you are moving from the Amazon Q Developer IDE plugin, copy the
aicommanderentry into Kiro'smcp.jsonand check its fields against Kiro's format, rather than re-creating it by hand.
FAQ
mcp.json: user scope ~/.kiro/settings/mcp.json or workspace scope .kiro/settings/mcp.json, with mcpServers.aicommander.url set to https://aicommander.dev/mcp. In the IDE, run Kiro: Open user MCP config (JSON) from the command palette. Prefer stdio? Use command npx with args -y @aicommander/mcp. The remote URL may ask for a one-time browser sign-in (OAuth); the npx bridge never does: it connects anonymously unless AICOMMANDER_TOKEN is set in its env. On the remote URL, use ?anonymous=1 for a code-only first hour. See /howto/ for the full matrix.mcp.json again.Next steps
Install the agent on a machine you control, add AI Commander to Kiro’s mcp.json (remote URL or npx -y @aicommander/mcp), then ask Kiro for a real task on a remote box.