Granted MCP
Discover grant opportunities across funders, with eligibility and deadline details.
Verdict
Common use cases
- Audit who accessed sensitive resources last week
- Request temporary elevated permissions for deployments
- Review compliance before quarterly security audits
- Debug why a teammate can't access a service
- Revoke stale access across multiple platforms
Integration
- Vendor
- Granted MCP
- Category
- other
- Auth
- NONE
- Composio slug
granted_mcp
Tools
Setup
Setup guide
- 11. Open your Switchy workspace and navigate to Settings > Integrations > MCP Servers. 2. Click 'Add MCP Server' and select Granted MCP from the list. 3. Since this MCP uses no authentication by default, you'll configure which access control systems it monitors — expect to provide API keys or service account credentials for AWS IAM, Okta, Google Workspace, or similar platforms during this step. 4. Save the configuration and return to any Project. 5. Type '@Granted' in the chat to confirm the connection works — ask it to list available commands or describe what access systems it can query. 6. Test a real workflow by asking '@Granted who has admin access to our production environment' or '@Granted request temporary read access to the billing dashboard for 2 hours'.
What teammates see: by default, memories from Granted MCP are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Audit Recent Access
@Granted show me everyone who accessed production databases or admin panels in the last 7 days, grouped by resource type
Request Temporary Access
@Granted I need read-only access to the billing API for the next 3 hours to debug an invoice discrepancy — can you grant that and notify my manager?
Compare Team Permissions
@Granted compare access levels for Alice and Bob — they're both on the platform team and should have the same permissions to staging and production
Find Stale Access
@Granted list all service accounts and user accounts that have admin access but haven't authenticated in the past 90 days
Explain Access Denial
@Granted why can't jordan@example.com deploy to the staging environment? Walk through the permission chain and show where access is denied
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@granted list all my AWS accounts and show which ones I currently have active credentials for
Based on Granted's configuration, you have access to 4 AWS accounts: prod-main (123456789012), staging-env (234567890123), dev-sandbox (345678901234), and analytics-prod (456789012345). Currently, you have active credentials for staging-env, which expire in 47 minutes. The other accounts require you to run 'assume' to generate temporary credentials.
This example shows Granted's core value: managing multiple AWS account credentials from a single interface. The MCP reads your local Granted configuration and credential state. No AWS API calls are made here — it's inspecting what's already on your machine.
@granted assume the prod-main account with read-only access and tell me what region it defaults to
Credentials assumed for prod-main (123456789012) with the ReadOnlyAccess role. Session valid for 1 hour. Default region is set to us-east-1 based on your Granted profile configuration. You can now run AWS CLI commands or use other MCPs that need AWS credentials without manually exporting environment variables.
This demonstrates Granted's assume workflow, which generates temporary STS credentials for a specific role. The MCP triggers the credential exchange and confirms the session parameters. Be aware this creates a time-limited session — you'll need to re-assume after expiry.
@granted check which accounts have credentials expiring in the next 30 minutes, then draft a Slack message reminding the team to refresh before the deploy window
Two accounts have credentials expiring soon: staging-env (expires in 22 minutes) and dev-sandbox (expires in 28 minutes). Here's a draft Slack message: 'Heads up team — staging and dev credentials expire in ~25 minutes. If you're planning to deploy during the 3pm window, run `assume staging-env` now to refresh your session. Let me know if anyone hits auth issues.' This gives the team a 5-minute buffer before the actual expiry.
This example pairs Granted's credential-state awareness with the AI's ability to synthesize context. The MCP provides the raw expiry data; the AI shapes it into a useful team communication. Useful for teams doing coordinated deploys across multiple AWS accounts.
Use-case deep-dives
When you need a quick snapshot of who can do what
A 6-person SaaS startup gets its first SOC 2 audit request and needs to document access controls across AWS, GitHub, and Slack. Granted MCP is built for this: it pulls permission data from connected systems and surfaces it in a single view during a Switchy session. Your auditor asks "who has admin on production?", you query the MCP, and you get an answer in seconds instead of spreadsheet archaeology. The trade-off: if your stack isn't supported by Granted's integrations, you're back to manual lookups. This works when you have 5-15 services and need occasional visibility, not continuous monitoring. If compliance is your day job, you want a dedicated tool. For the rest of us, this MCP turns a 3-hour access audit into a 20-minute Switchy conversation.
Granted MCP speeds up new-hire setup for small teams
Your design team hires a contractor and you need to grant Figma, Linear, and Vercel access by Monday. Granted MCP lets you check current permissions and request changes without leaving Switchy's chat interface. You ask "what access does the design role usually get?", compare it to what the contractor needs, and file the requests in one session. The boundary: this assumes your identity provider or admin tooling already integrates with Granted. If you're stitching together OAuth apps with no central directory, the MCP can't help. This is a win for teams under 20 people who provision access weekly, not daily. Larger orgs need workflow automation, not a chat interface. For early-stage teams, it collapses 6 Slack threads into one Switchy transcript.
When someone leaves and you need a clean deprovisioning checklist
A contractor wraps their engagement on Friday and you need to revoke access across 8 services before Monday. Granted MCP queries what the person currently has, surfaces it in Switchy, and you work through the list with your ops lead in real time. You catch the Stripe admin role they forgot they had. The constraint: this only works if Granted can read your access graph. If half your tools are unconnected SaaS with separate logins, you're still doing manual cleanup. This scenario wins when you have 10-30 connected services and offboard people monthly. If turnover is higher, automate it with an identity platform. For occasional offboarding at small scale, this MCP turns a risky checklist into a verified runbook you execute in Switchy.
Frequently asked
What does the Granted MCP integration do in Switchy?
Granted MCP connects your team's access request workflows to Switchy's AI workspace. It lets agents query who has access to what resources, check pending approval requests, and surface access patterns without leaving the conversation. This is useful when debugging permission issues or auditing who can reach production systems during incident response.
Does Granted MCP require authentication to connect?
No. This MCP uses no authentication, which typically means it either reads from local configuration files or connects to a self-hosted Granted instance on your network. You won't go through OAuth or enter API keys. Check that your Granted deployment is reachable from wherever Switchy runs your agents.
Can Granted MCP approve or revoke access requests automatically?
That depends on which tools the MCP exposes, but most access control integrations are read-only by design for safety. Expect to query current grants and request status, not mutate permissions. If you need write operations, you'll likely handle approvals in Granted's own interface and use the MCP for visibility only.
Why use Granted MCP instead of just checking the Granted dashboard?
The MCP brings access data into the same workspace where your team is already troubleshooting or planning deployments. An agent can answer "does Alice have prod database access?" inline, without context-switching to another tab. It's faster for ad-hoc questions and keeps the audit trail in your Switchy conversation history.
Who on the team should set up the Granted MCP connection?
Whoever manages your Granted deployment or has visibility into its API endpoints. Since there's no OAuth flow, setup is usually pointing the MCP at the right hostname or config file path. Your security or platform team will know whether Granted is network-accessible from Switchy's environment.