walkindb-mcp
v0.1.1
Published
Model Context Protocol server for walkindb — gives Claude, Cursor, Zed, and other MCP clients access to the database for agents: a private SQL database with a 10-minute TTL. No signup, no API key, no credit card.
Maintainers
Readme
walkindb-mcp
Model Context Protocol server for walkindb — the database for agents.
Adds two tools to any MCP-compatible client (Claude Code, Claude Desktop, Cursor, Zed, Continue, ...):
walkindb_execute(sql)— run one SQL statement against a private, ephemeral SQLite database. The first call provisions a new walk-in; subsequent calls in the same MCP session reuse the same database until its 10-minute TTL expires.walkindb_reset()— forget the current session so the nextwalkindb_executecall provisions a fresh walk-in.
No signup, no API key, no credit card. The underlying walk-in database lives for ~10 minutes then disappears — it's for agent scratch state, not anything durable.
Install
No manual install required — MCP clients typically launch the server via npx:
npx walkindb-mcpWire it into your client
Claude Desktop
Edit your claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json, Windows: %APPDATA%\Claude\claude_desktop_config.json):
{
"mcpServers": {
"walkindb": {
"command": "npx",
"args": ["-y", "walkindb-mcp"]
}
}
}Restart Claude Desktop. Claude now has walkindb_execute and walkindb_reset tools.
Claude Code
claude mcp add walkindb -- npx -y walkindb-mcpOr edit .claude/mcp.json in your project:
{
"mcpServers": {
"walkindb": {
"command": "npx",
"args": ["-y", "walkindb-mcp"]
}
}
}Cursor
Edit ~/.cursor/mcp.json:
{
"mcpServers": {
"walkindb": {
"command": "npx",
"args": ["-y", "walkindb-mcp"]
}
}
}Zed
Add to ~/.config/zed/settings.json under "context_servers":
{
"context_servers": {
"walkindb": {
"command": {
"path": "npx",
"args": ["-y", "walkindb-mcp"]
}
}
}
}Continue
Add to ~/.continue/config.json under "experimental.modelContextProtocolServers":
{
"experimental": {
"modelContextProtocolServers": [
{
"transport": {
"type": "stdio",
"command": "npx",
"args": ["-y", "walkindb-mcp"]
}
}
]
}
}Any other MCP client
walkindb-mcp speaks the standard Model Context Protocol over stdio. If your client can spawn a command and talk JSON-RPC on stdin/stdout, it can use walkindb-mcp.
Environment variables
| Variable | Default | Purpose |
|---|---|---|
| WALKINDB_BASE_URL | https://api.walkindb.com | Override for self-hosted deployments |
| WALKINDB_TIMEOUT_MS | 10000 | Per-request timeout in milliseconds |
What the tools return
walkindb_execute
On success, returns a JSON text block:
{
"columns": ["id", "body"],
"rows": [[1, "hello"]],
"rows_affected": 0,
"truncated": false,
"session_established": "wkn_AZ159u9PdmS97ks7...",
"expires_at": 1775868670
}On error (bad SQL, forbidden keyword, rate limit, etc.), returns a text block with isError: true:
walkindb error 400: forbidden sql keyword: ATTACHwalkindb_reset
Returns a short confirmation. Does not delete the server-side walk-in — that's still deleted by the TTL sweeper in its normal window.
What walkindb is for
- Agent scratchpad memory across tool calls in a single run
- RAG chunk staging and re-ranking with SQL
- One-shot analyses of CSVs and other tabular data
- Throwaway SQL workloads where you want real joins, window functions, CTEs
What walkindb is NOT for
- Anything durable — walk-ins disappear after ~10 minutes
- PII or regulated data
- Credentials, secrets, or anything you can't afford to leak
- Large datasets (10 MB per instance)
See the Acceptable Use Policy for the full list.
Limits
- 10 MB per walk-in database
- 8 KB per SQL statement
- 2-second wall-clock timeout per statement
- 500 000 VDBE operations per statement
- 60 requests / minute / IP, 10 new instance creations / minute / IP
Docs
Apache 2.0. Source at github.com/walkindb/walkindb.
