Back to blog
New Integration

Cursor Cloud Agents Meet LeemerChat

Launch, monitor, and manage Cursor cloud agents from inside your chat. Send follow-up instructions, check status, create PRs — all with natural language.

10 APIs
Endpoints
Basic Auth
Auth
Chat Tools
Integration
BYOK
Access

Why Cursor Cloud Agents + LeemerChat?

We already have PowerCode for reading your codebase and Blackbox Cloud for multi-agent dispatch. Cursor Cloud Agents adds a third superpower: iterative, conversational coding agents that you can refine with follow-up instructions — all from the same chat interface.

Natural Language → Code Changes

Describe what you want in plain English inside the chat. The AI launches a Cursor cloud agent, monitors progress, and reports back — no context switching to another platform.

Iterative Follow-ups

Unlike one-shot agents, Cursor supports full conversation threads. Send follow-up instructions to refine the work, add tests, or adjust the implementation — all from chat.

Full Lifecycle in Chat

Launch agents, check status, send follow-ups, stop, resume, delete — all without leaving the conversation. Your chat history becomes a log of every agent dispatched.

Zero Lock-in

Bring your own Cursor API key. The integration is fully optional — if you don't add a key, the AI has no idea these tools exist. Remove the key any time to turn it off.

How It Works: Chat Tools

The integration adds seven tools to LeemerChat's AI assistant. These tools are only visible when you've added your Cursor API key in Settings → API Keys. No key, no tools — the AI has zero knowledge of Cursor otherwise.

cursor_launch_agent

Launch a new cloud agent on any GitHub repository. Supports custom models, auto-PR creation, and custom branch names.

You say:

"Use Cursor to add rate limiting to github.com/myorg/api"

cursor_check_agent

Check the current status, summary, branch, and PR URL of a running or completed Cursor agent.

You say:

"What's the status of my Cursor agent?"

cursor_stop_agent

Stop a running agent. The agent is paused but not deleted — send a follow-up to resume.

You say:

"Stop the Cursor agent"

cursor_followup_agent

Send a follow-up instruction to an existing agent. If the agent was stopped, this resumes it.

You say:

"Tell the Cursor agent to also add unit tests"

cursor_list_agents

List all your Cursor cloud agents with their names, statuses, repos, and PR links.

You say:

"Show me all my Cursor agents"

cursor_delete_agent

Permanently delete a cloud agent. This action cannot be undone.

You say:

"Delete my old Cursor agent bc_abc123"

cursor_list_repos

List GitHub repositories accessible to your Cursor account to find the right repo before launching an agent.

You say:

"List my repos on Cursor"

Example: Iterative Agent Workflow

Here's what a typical Cursor agent workflow looks like, entirely inside the chat:

1
You

"Use Cursor to add authentication middleware to github.com/myorg/api on the main branch, and auto-create a PR"

2
LeemerChat AI

AI calls cursor_launch_agent with repository, ref, and autoCreatePr. Returns agentId: bc_abc123.

3
You

"How's the agent doing?"

4
LeemerChat AI

AI calls cursor_check_agent(bc_abc123). Reports: status "RUNNING", branch: cursor/add-auth-1234.

5
You

"Also add rate limiting to the auth endpoints"

6
LeemerChat AI

AI calls cursor_followup_agent(bc_abc123, prompt). The agent continues working with the new instruction.

7
LeemerChat AI

Agent finishes. Status: "FINISHED". PR created at github.com/myorg/api/pull/42. Summary: "Added JWT auth middleware with rate limiting."

PowerCode vs Blackbox Cloud vs Cursor Cloud

Three tools in LeemerChat. They solve different problems and work great together.

Feature PowerCode Blackbox Cloud Cursor Cloud
Where it runsInside LeemerChat's chat interfaceRemote Blackbox Cloud sandboxesCursor Cloud infrastructure
How it worksBrowse & read files via GitHub APIClone repo, make changes, commit & pushClone repo, make changes, create branch & optional PR
OutputIn-chat code suggestions & analysisNew branch with committed changes + PRNew branch with changes + optional auto-PR
Agent countSingle model at a timeUp to 4 agents simultaneously (multi-launch)Single agent with follow-up conversation support
Best forCode exploration, Q&A, quick fixesFull implementations, refactors, featuresIterative coding tasks with follow-up refinement
Cost modelIncluded in LeemerChat planCredits on your Blackbox account (BYOK)Billed via Cursor subscription (BYOK)

Getting Started

1

Get your Cursor API key

Visit cursor.com/settings (Cursor Dashboard) and generate an API key. The key uses Basic Authentication to authenticate your requests.

2

Add the key in LeemerChat

Open Settings, then API Keys, and paste your Cursor API key. Once saved, the seven Cursor tools become available to the AI assistant.

3

Start launching agents

Just ask in chat: "Use Cursor to refactor the auth module in github.com/myorg/api and create a PR". The AI handles the rest — launching, monitoring, and reporting.

Privacy & Security

Your Cursor API key is stored only in your browser's localStorage — never on our servers.

API calls to Cursor go through our authenticated server proxy — your key never appears in browser network logs.

If you remove the key, all Cursor tools disappear from the AI instantly.

We do not store, log, or share any data from your Cursor agents. Usage is billed via your Cursor subscription.

Start Launching Cursor Agents Today

Add your Cursor API key, pick a repo, describe what you want, and let the agent handle the rest — all from inside LeemerChat.