Cursor
Cursor is an AI-powered code editor built on VSCode. The API provides access to Cloud Agents, Admin, Analytics, and AI Code Tracking features.
Verdict
Common use cases
- Review AI pair-programming sessions from Cursor
- Audit code changes suggested by agents
- Check which models your team can access
- List repositories synced to Cursor accounts
- Verify API key ownership and creation dates
Integration
- Vendor
- Cursor
- Category
- developer-tools
- Auth
- API_KEY
- Tools
- 5
- Composio slug
cursor
Tools
- Get Agent Conversation
Tool to retrieve the conversation history for a specific cloud agent. Use when you need to view the messages exchanged between the user and the agent. Returns error with status 404 if the agent is not found or access is denied.
- Get API Key Info
Tool to retrieve API key information including the key name, creation date, and owner email. Use when you need to verify API key details or validate authentication.
- List Agents
Tool to retrieve a paginated list of all Cursor Cloud agents. Use when you need to view existing agents, their status, or iterate through agents with pagination.
- List Available Models
Tool to retrieve the list of available AI models in Cursor. Use when you need to discover which models are available for API requests.
- List GitHub Repositories
Tool to list GitHub repositories accessible to the authenticated user. Use when you need to retrieve all repositories the user has access to through their Cursor account.
Setup
Setup guide
- 11. In Switchy, navigate to Settings > Integrations and select Cursor from the developer tools section. 2. Click 'Connect Cursor' to open the authentication modal. 3. Log into your Cursor account and generate an API key from your account settings under 'API Access'. 4. Copy the API key and paste it into the Switchy modal, then click 'Authorize'. 5. Switchy confirms the connection by calling the Get API Key Info tool to verify your key and display the owner email. 6. Open any Space and type '@Cursor list my agents' to test — you should see a paginated list of your Cursor Cloud agents. 7. To retrieve a specific conversation, use '@Cursor get agent conversation [agent_id]' with an ID from the list.
What teammates see: by default, memories from Cursor are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
List All Agents
@Cursor list all my Cursor Cloud agents with their current status and when they were createdOpen in a Space →
Fetch Agent Transcript
@Cursor get the conversation history for agent ID [paste_agent_id_here] and summarize the key code changes discussedOpen in a Space →
Check Available Models
@Cursor list all available AI models in my Cursor workspace and highlight which ones are best for TypeScript refactoringOpen in a Space →
Verify API Key Details
@Cursor retrieve my API key information including the owner email and creation timestampOpen in a Space →
List Linked Repositories
@Cursor show me all GitHub repositories accessible through my Cursor account and flag any that are archivedOpen in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@cursor list all my cloud agents and show me which ones are currently active or have recent conversations
Found 4 Cursor Cloud agents in your account: 1. **feature-auth-refactor** (active) — Last conversation 2 hours ago, 23 messages exchanged 2. **bug-payment-flow** (idle) — Last conversation 3 days ago, 15 messages 3. **docs-api-update** (active) — Last conversation 45 minutes ago, 8 messages 4. **test-ci-pipeline** (idle) — Last conversation 1 week ago, 31 messages This is an illustrative example showing how the MCP surfaces agent metadata. Actual response includes agent IDs, creation timestamps, and conversation counts.
Demonstrates the List Agents tool with pagination awareness. Requires a valid Cursor API key with read access to cloud agents. Useful for auditing which coding tasks have active AI assistance or reviewing past agent sessions before resuming work.
@cursor pull the full conversation history from my 'feature-auth-refactor' agent so I can review what we decided about the OAuth flow
Retrieved 23 messages from agent **feature-auth-refactor**: **User (2h ago):** Should we use PKCE for the mobile OAuth flow? **Agent:** Yes, PKCE is recommended for mobile clients to prevent authorization code interception... **User (2h ago):** What about refresh token rotation? **Agent:** Implementing refresh token rotation adds security by limiting token lifetime... [18 more messages showing code suggestions, security trade-offs, and implementation decisions] This illustrative example shows how conversation retrieval works. Real output includes full message content, timestamps, and code blocks.
Uses the Get Agent Conversation tool to fetch historical context from a specific cloud agent. Returns a 404 error if the agent ID doesn't exist or isn't accessible to your API key. Valuable for resuming interrupted coding sessions or documenting architectural decisions made during pair-programming with Cursor AI.
@cursor check which AI models I have access to, then summarize the conversation from my most recent agent to draft a quick status update for the team
**Available Models:** claude-3-5-sonnet-20241022, gpt-4o, gpt-4o-mini, o1-preview, o1-mini **Status Update from 'docs-api-update' agent (last 45 min):** Completed API documentation refresh for v2.3 endpoints. Agent helped restructure authentication examples to show both API key and OAuth flows. Identified 3 deprecated parameters still referenced in old docs—flagged for removal. Next: review rate limit section for accuracy. This illustrative example combines model discovery with conversation synthesis. Actual output depends on your agent's recent activity and available model list.
Chains two MCP tools (List Available Models + Get Agent Conversation) with AI reasoning to produce a human-readable summary. Demonstrates how Cursor MCP integrates with Switchy's synthesis capabilities. Requires API key access to both models metadata and agent history—useful for standups or handoff documentation.
Use-case deep-dives
When you need to reconstruct what the AI actually did
A 6-person backend team ships a bug that traces back to an AI-generated migration script. The tech lead needs to see exactly what the agent was told, what context it had, and how it responded. The Cursor MCP's Get Agent Conversation tool pulls the full exchange in under 10 seconds—no digging through local IDE logs or Slack threads. This works if your team runs Cursor Cloud agents and keeps API keys scoped to the project. If you're using local-only Cursor sessions, this MCP can't help; the conversation history lives on individual machines. The buying call: if your team uses Cursor Cloud for any production work, wire this MCP into your incident-response runbook before you need it.
Right-sizing AI spend across a distributed team
A 12-person startup uses Cursor across three time zones. The CTO notices the monthly AI bill jumped 40% but can't tell which models are burning tokens. The Cursor MCP's List Available Models tool surfaces the full catalog, and Get API Key Info shows which keys belong to which engineers. The team writes a quick script to log model usage per key, then sets a policy: junior devs default to the cheaper model unless they opt up. This scenario assumes your team uses API keys (not OAuth) and that you're willing to run a lightweight usage-tracking layer. If you have fewer than 5 engineers or a flat AI budget, the overhead isn't worth it. The call: adopt this MCP when your team crosses 8 people or $500/month in Cursor spend.
Verifying new hires can reach the right codebases
A 4-person agency hires a contractor for a two-week sprint. The PM wants to confirm the contractor's Cursor account can see the client repo before kickoff. The Cursor MCP's List GitHub Repositories tool returns the full list of accessible repos in one call—no need to ask the contractor to screenshot their IDE or manually check GitHub org settings. This works if your team uses GitHub (not GitLab or Bitbucket) and if the contractor has already linked their Cursor account to GitHub. If your repos live elsewhere, this tool returns an empty list. The threshold: if you onboard more than one external dev per quarter, this MCP saves 15 minutes of back-and-forth every time.
Frequently asked
What does the Cursor MCP do in Switchy?
It connects your team's Cursor AI editor to Switchy's workspace. You can list cloud agents, pull conversation histories, check which models are available, and see GitHub repos linked to your Cursor account. Useful for auditing what your developers are asking Cursor's AI or syncing agent context into other tools.
Do I need a Cursor Pro account to use this MCP?
Yes. The MCP authenticates with a Cursor API key, which requires a paid Cursor subscription. You generate the key in your Cursor account settings, then paste it into Switchy. Free Cursor accounts don't expose an API, so the integration won't work without upgrading.
Can the Cursor MCP create new agents or send messages?
No. It's read-only. You can list agents, retrieve conversation histories, and check available models, but you can't start new chats or inject messages into existing ones. If you need to automate agent creation, use Cursor's CLI or web interface directly.
Why use this instead of just opening Cursor?
The MCP lets you pull Cursor agent data into Switchy workflows without switching apps. For example, you can log all AI conversations to Notion, compare model usage across your team, or trigger alerts when an agent references a specific repo. Cursor's UI doesn't expose this programmatically.
Who on the team should connect the Cursor MCP?
Whoever owns the Cursor account you want to monitor. Each API key is scoped to one user's agents and repos. If you want visibility across multiple developers, each person needs to connect their own Cursor account to Switchy as a separate integration.