Skip to content

feat: add MCP server for AI integration#524

Closed
joshrotenberg wants to merge 1 commit intomainfrom
feat/mcp-server
Closed

feat: add MCP server for AI integration#524
joshrotenberg wants to merge 1 commit intomainfrom
feat/mcp-server

Conversation

@joshrotenberg
Copy link
Copy Markdown
Collaborator

Summary

Add a new redisctl-mcp crate that provides a Model Context Protocol (MCP) server enabling AI assistants like Claude to manage Redis deployments.

Closes #475

What's Included

Phase 1: Read-Only Tools + Database Create

Cloud Tools (7 read-only):

  • cloud_account_get - Get account information
  • cloud_subscriptions_list - List all subscriptions
  • cloud_subscription_get - Get subscription details
  • cloud_databases_list - List databases in a subscription
  • cloud_database_get - Get database details
  • cloud_tasks_list - List recent async tasks
  • cloud_task_get - Get task status

Enterprise Tools (10 read-only + 1 write):

  • enterprise_cluster_get - Get cluster information
  • enterprise_nodes_list - List all cluster nodes
  • enterprise_node_get - Get node details
  • enterprise_databases_list - List all databases
  • enterprise_database_get - Get database details
  • enterprise_database_stats - Get database statistics
  • enterprise_shards_list - List all shards
  • enterprise_alerts_list - List active alerts
  • enterprise_logs_get - Get cluster event logs
  • enterprise_license_get - Get license information
  • enterprise_database_create - Create a new database (write)

Usage

# Start MCP server (read-only mode, default)
redisctl -p my-profile mcp serve

# Enable write operations
redisctl -p my-profile mcp serve --allow-writes

# List available tools
redisctl mcp tools

Claude Desktop Configuration

{
  "mcpServers": {
    "redisctl": {
      "command": "/path/to/redisctl",
      "args": ["-p", "my-profile", "mcp", "serve"]
    }
  }
}

Other Changes

  • Documentation: Added docs/src/integrations/mcp.md with full usage guide
  • Bug fix: Changed search and timeseries fields in redis-enterprise crate from Option<bool> to Option<Value> to handle API responses that return objects

Testing

Tested with live Redis Enterprise cluster:

  • All read tools work correctly
  • Database creation works via MCP
  • Read-only mode properly blocks write operations

Next Steps (Future PRs)

  • Add more write operations (delete, update)
  • Add Cloud write operations
  • Add resource support for prompts/context

Add a new redisctl-mcp crate that provides a Model Context Protocol (MCP)
server enabling AI assistants like Claude to manage Redis deployments.

Phase 1 implements 18 tools:

Cloud Tools (7 read-only):
- cloud_account_get, cloud_subscriptions_list, cloud_subscription_get
- cloud_databases_list, cloud_database_get
- cloud_tasks_list, cloud_task_get

Enterprise Tools (10 read-only + 1 write):
- enterprise_cluster_get, enterprise_nodes_list, enterprise_node_get
- enterprise_databases_list, enterprise_database_get, enterprise_database_stats
- enterprise_shards_list, enterprise_alerts_list, enterprise_logs_get
- enterprise_license_get
- enterprise_database_create (write, requires --allow-writes)

Usage:
  redisctl -p my-profile mcp serve              # read-only mode
  redisctl -p my-profile mcp serve --allow-writes  # enable writes
  redisctl mcp tools                            # list available tools

Also includes:
- Documentation for MCP integration (docs/src/integrations/mcp.md)
- Fix for search/timeseries field types in redis-enterprise crate
@joshrotenberg
Copy link
Copy Markdown
Collaborator Author

Superseded by #531 which includes all MCP functionality

@joshrotenberg joshrotenberg deleted the feat/mcp-server branch March 6, 2026 03:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

feat: Add MCP (Model Context Protocol) server for AI integration

1 participant