cloud-agent-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@cloud-agent-mcplaunch a cloud agent to fix the lint failures and open a PR"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
@mciver87/cloud-agent-wrapper
CLI + MCP server for launching, resuming, and managing Cursor Cloud Agents from any terminal or MCP-capable agent.
Backed by @cursor/sdk. One shared core, two front-ends:
CLI (
cloud-agent new|list|get|resume|cancel) — for humans, shell scripts, CI, or any agent that can shell out.MCP server (
cloud-agent-mcp) — exposes the same operations as native tools to Cursor Desktop,cursor-agentCLI, Claude Desktop, Codex, and any other MCP-capable client.
Install
npm install -g @mciver87/cloud-agent-wrapper
export CURSOR_API_KEY="crsr_..." # from https://cursor.com/t/<team>/apiRelated MCP server: cursor-sdk-mcp
CLI usage
# Launch — auto-detects repo from `git remote get-url origin`
cloud-agent new "fix the lint failures and open a PR"
# Explicit repo + model + auto-PR
cloud-agent new --repo foo/bar --model opus-5.5 --pr "rebase onto main"
# List recent cloud agents
cloud-agent list
cloud-agent list --json --limit 50
# Fetch metadata
cloud-agent get bc-abc123
# Follow-up on an existing agent
cloud-agent resume bc-abc123 "also update the changelog"
# Cancel a running agent
cloud-agent cancel bc-abc123Exit codes
0— run finished successfully1— startup / config / auth failure (couldn't even begin)2— run executed but completed with an error status
JSON mode
Add --json to any command for machine-readable output. For new and resume, each line of stdout is one SDKMessage JSON object; for list/get/cancel, a single JSON object.
MCP server usage
Register in your client's mcp.json. Example for Cursor (~/.cursor/mcp.json):
{
"mcpServers": {
"cloud-agent": {
"command": "cloud-agent-mcp",
"env": {
"CURSOR_API_KEY": "crsr_..."
}
}
}
}After restarting the client, these tools become callable natively:
Tool | What |
| New agent; returns ID, URL, and final assistant text |
| List recent agents on your account |
| Metadata for one agent |
| Follow-up prompt on existing agent |
| Stop a running agent |
Clients that respect inputSchema get full argument validation (Zod schemas). Clients that don't still see rich descriptions in the tool list.
Model shorthands
Any of these --model values map to full IDs:
Shorthand | Resolves to |
|
|
|
|
|
|
|
|
Pass any other string through verbatim. If omitted, your team's default model is used.
Repo resolution
If --repo is omitted, cloud-agent runs git remote get-url origin in the current directory and parses owner/name from the URL (both git@github.com:owner/name and https://github.com/owner/name[.git] forms work). Pass --repo owner/name or --repo https://github.com/owner/name to override.
API key
Required: CURSOR_API_KEY env var. Generate at cursor.com/t/<team>/api. For long-term use, add to ~/.zshenv (chmod 600) so every shell + subprocess inherits it.
Service-account keys work the same way; the Cursor backend decides whether PRs open as the key owner or the Cursor GitHub App.
Development
git clone https://github.com/mciver87/cloud-agent-wrapper
cd cloud-agent-wrapper
npm install
npm run build
npm run typecheckSource layout:
src/core.ts— SDK wrapper (launch/list/get/resume/cancel). All SDK knowledge lives here.src/cli.ts— commander-based CLI, exits with 0/1/2 per the SDK skill convention.src/mcp.ts— MCP server, same operations as native tools.bin/cloud-agent.mjs+bin/cloud-agent-mcp.mjs— ESM shebang shims.
License
MIT
This server cannot be deployed
Maintenance
Related MCP Connectors
Your org's AI agents, tasks, runs, search, and brain files as MCP tools and resources.
MCP server for your apps' tools and custom tools, plus hosted AI agents and approval-gated workflows
Discover and call AI agents via MCP. Supports A2A agents and platform agents with async tasks.
MCP facade over the Nebelus Construction API. ~48 tools give full agent build parity: create/update/probe agents, edit graphs, attach knowledge and vector stores, wire connectors, set governance policies and locked guardrails, enable grounding-trace, and read deployment wiring. Purpose-built for regulated industries: data residency is enforced per region (EU / GCC-KSA), with PII controls and an audit trail. Agents are created as drafts — no deploy tool is exposed over MCP by design; publishing happens in the Nebelus console.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables programmatic control of Cursor's cloud-based AI agents for automated code generation and repository management via standardized MCP tools.16 npm6MIT
- AlicenseAqualityDmaintenanceEnables MCP clients to invoke Cursor SDK's agent runtime, run coding agents, list models, and continue conversations.41MIT
- AlicenseNot gradedqualityDmaintenanceEnables Claude to launch and manage Cursor Cloud Agents on GitHub repos, including polling status, sending follow-ups, and reading conversations/PRs.16 npmMIT
- AlicenseAqualityCmaintenanceEnables AI assistants and other MCP clients to launch and steer Cursor Cloud Agents on repositories or as repo-less research tasks, including model selection, mid-run follow-ups, cancellation, result retrieval, and run listing. Also exposes authentication checks and per-agent or per-run token usage and cost reporting.9MIT