Skip to main content
The redis MCP adapter connects to Redis with an address (host:port), optional ACL username/password, and a logical database index. It provides key-value helpers (GET, SET, INCR, TTL management) plus a generic redis_run_command escape hatch for any Redis command. Use it for caching, rate limiting, session state, distributed counters, idempotency guards, and lightweight coordination alongside Postgres or MySQL backends.

Prerequisites

  • A Redis server (standalone, Sentinel, or managed — ElastiCache, Memorystore, etc.) reachable from AgentRuntime
  • Connection address as host:port and optional ACL username/password
  • Logical database index if not using the default (0)
  • TLS requirements understood if your provider requires encrypted connections (configure at the network/proxy layer)
  • project_contributor access

Connect in AgentRuntime

1

Create a connection

Go to Connections and click New custom connection to create a connection with your Redis settings:
  • Addr — server address as host:port (e.g. redis.example.com:6379)
  • Password — Redis password or ACL secret (omit for unauthenticated Redis)
  • Username — ACL username when required (omit for password-only auth)
  • DB — logical database index (default 0)
2

Add redis MCP instance

Go to MCP, click Add instance for redis, wire your connection on Instance config, set the profile active, and save.
3

Test connectivity

Add a workflow with mcp_callredis_run_command with command: "PING". Run it to confirm the server responds before production use.
4

Verify key operations (optional)

Run mcp_callredis_set with a test key and short TTL, then redis_get to confirm read/write access on the selected DB index.
See Authentication for binding connections to MCP instances.

What you can build

  • Rate-limited API callsredis_incr on a per-tenant key → branch workflow when count exceeds threshold.
  • Idempotency guardredis_exists on a webhook event ID → skip duplicate processing with redis_set + TTL.
  • Cache warming — Scheduled fetch from Postgresredis_set with ttl_seconds for fast Autopilot lookups.
  • Distributed counterredis_incr / redis_decr for usage metering → sync totals to billing on interval.

Tools

Example

Cache an API response for one hour:

Configuration

Troubleshooting