Redis MCP Server
Read-only Redis and Valkey explorer with safe key scans and a big-key advisor
- Pricing
- Open source
- Platforms
- CLI, macOS, Linux
- Runs
- Locally
- Version
- 0.5.8
- Listing updated
- 2 days ago
About Redis MCP Server
Redis MCP Server is an open-source MCP server for databases. It runs locally on your machine and is started by your AI client.
Safety on a live instance is the main design point. Key enumeration always uses cursor-based SCAN with a bounded count and iteration cap, never the O(N) KEYS command that can block Redis's single-threaded loop, so the README says it is fine to point at a production instance with millions of keys. Tools run a curated read-only command allowlist, and writes such as SET, DEL, EXPIRE and HSET only work if ALLOW_WRITES=1 is set. EVAL, FUNCTION, SCRIPT, MULTI, MONITOR, SHUTDOWN and cluster commands are never exposed, even with writes on.
redis_health rolls INFO, DBSIZE and recent SLOWLOG into one report covering memory pressure, eviction policy, hit rate, ops per second, persistence, replication role and per-database key counts. redis_advisor acts as a lint pass, flagging big keys, missing TTLs, eviction pressure including the risky noeviction with no-TTL combination, and fork-latency risk, each with a severity and a fix. redis_key_info reports type, TTL, encoding, memory footprint and idle time without reading the value, and redis_get windows collection reads so a huge list cannot flood the model's context.
It works with Redis 6 and later and with Valkey, and a field the server does not report shows as null. Set REDIS_URL, and note that a password-only URL needs the form redis://:password@host:6379. It ships as one bundled file with no runtime dependencies, is MIT licensed and is built by Yaw Labs.
Install
Run in a terminal
claude mcp add redis-mcp -- npx -y @yawlabs/redis-mcpAdd to claude_desktop_config.json
{
"mcpServers": {
"redis-mcp": {
"command": "npx",
"args": [
"-y",
"@yawlabs/redis-mcp"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"redis-mcp": {
"command": "npx",
"args": [
"-y",
"@yawlabs/redis-mcp"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"redis-mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@yawlabs/redis-mcp"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"redis-mcp": {
"command": "npx",
"args": [
"-y",
"@yawlabs/redis-mcp"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"redis-mcp": {
"command": "npx",
"args": [
"-y",
"@yawlabs/redis-mcp"
]
}
}
}Tools this server provides (7)
Taken from the project documentation; the running server may differ.
- redis_scan
- Enumerate keys with cursor-based SCAN, with optional glob match and type filter.
- redis_key_info
- Inspect one key's type, TTL, encoding, memory and idle time without reading its value.
- redis_health
- Summarise INFO, DBSIZE and recent SLOWLOG into one health report.
- redis_advisor
- Report big keys, missing TTLs, eviction pressure and fork-latency risk with fixes.
- redis_slowlog
- Return recent slow-log entries with command, microseconds, timestamp and client.
- redis_get
- Read a value by type with windowed collection reads.
- redis_command
- Run allowed commands; mutating ones need ALLOW_WRITES=1.
What it does
- SCAN-based key explorer
- Pages through keys with a resumable cursor, glob match and type filter, capped at REDIS_MAX_KEYS.
- One-call health report
- Summarises memory, eviction, hit rate, persistence, replication and slow commands.
- DBA advisor
- Finds big keys, missing TTLs, eviction risks and fork-latency risk with suggested fixes.
- Key inspection
- Reads type, TTL, encoding, memory and idle time without fetching the value.
- Windowed reads
- Type-aware reads cap collection output to protect model context.
- Opt-in writes
- Mutating commands need ALLOW_WRITES=1, and arbitrary execution stays blocked.
Good for
- Find out what is stored in an unfamiliar Redis instance
- Check a production cache for big keys and missing expiry
- Review recent slowlog entries after a latency spike
- Run a quick health check before a deploy
Pricing
Open source
MIT licensed; free to use
Prices change. Check yaw.sh before you buy.
Redis MCP Server FAQ
Is it safe on production?
The README says so: it is read-only by default and enumerates keys with SCAN, never KEYS.
How do I enable writes?
Set ALLOW_WRITES=1; scoping that to dev and test instances is advised, and arbitrary-execution commands stay blocked.
Does it work with Valkey?
Yes, it works with Redis 6 and later and Valkey; some health fields depend on what INFO exposes.
What does it not cover?
EVAL, scripts, pub/sub, MONITOR, cluster management and replication control; the README points to redis-cli for those.
Is Redis MCP Server free?
Yes. Redis MCP Server is open source, so you can use and self-host it for free.
How do I add Redis MCP Server to Claude Code?
Run this command in your terminal: claude mcp add redis-mcp -- npx -y @yawlabs/redis-mcp
What are the best alternatives to Redis MCP Server?
The closest alternatives to Redis MCP Server are Midplane, PlatypusDB, Lorebase, Sovaria and model2data studio.
Alternatives to Redis MCP Server

Midplane
MCP serverPostgres gateway for AI agents with table grants, column masking and approval for writes
Runs locally- Open source
- 5 GitHub stars
- Databases

PlatypusDB
MCP serverGives MCP clients access to your PlatypusDB database connections, scoped per agent
Runs locally- Open source
- Databases

Lorebase
MCP serverLets agents read and change a governed, versioned Postgres knowledge graph
Runs locally- Open source
- Knowledge and memory
Sovaria
MCP serverManage a Sovaria backend (Postgres, auth, storage, functions) from an AI assistant
Not checked yet- 4 tools
- Remote
- Freemium
- Databases

model2data studio
MCP serverGenerate realistic connected test data, dbt projects and dashboards from a schema
- Remote
- Freemium
- Databases

semlayer
MCP serverInfers a semantic layer from your data warehouse and serves it to agents
Runs locally- Free
- Databases