About mcp-debezium
mcp-debezium gives MCP clients monitoring and management tools for Debezium CDC connectors through the Kafka Connect REST API. It is for teams maintaining change-data-capture pipelines, with read-only defaults, connector scoping, credential redaction, and guarded deletion.
Key features
- 01Inspect cluster, plugins, and connector status
- 02Read connector configs and topics
- 03Create, update, and restart connectors
- 04Pause or resume connectors
- 05Redact credentials and guard deletion
Use cases
- 01Find failed connectors or tasks
- 02Review connector config with secrets redacted
- 03Restart a failed task
- 04Restrict the connectors available for management
MCP support
As an MCP server, it exposes Debezium connector monitoring and management tools to MCP clients through the Kafka Connect REST API. Access modes, connector allowlists, credential redaction, deletion gates, and audit logs govern operations.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y @dockndevai/mcp-debeziumCONNECT_URLBase URL of the Kafka Connect REST API.DEBEZIUM_MODEAccess mode: read-only | read-write | admin.Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
7Saved tool list · updated 2w ago
This tool takes no parameters.
Details
From the project READMEmcp-debezium
A Model Context Protocol server for Debezium (via the Kafka Connect REST API). It lets an MCP-capable client (Claude Desktop, Claude Code, etc.) monitor and manage change-data-capture connectors — status, config, restarts, lifecycle — with behaviour controlled entirely by flags.
Safe by default: it starts read-only, can be scoped to an allowlist of connectors, protects critical connectors from mutation, redacts credentials in connector configs, and gates deletion behind an explicit opt-in.
Features
- Monitoring — worker/cluster info, connector list, plugins, per-connector config, status (with failed-task traces), and topics.
- Management — create/update, pause/resume, restart connector or a single task.
- Lifecycle — delete connectors (admin).
- Access modes —
read-only→read-write→admin, layered so a mode never exposes tools above its level. - Credential redaction —
database.passwordand other secret-shaped config values are replaced with***REDACTED***before results are returned. - Security flags — connector allowlist, protected connectors, delete gating, dry-run, and JSON audit logging (see below).
Security model
| Concern | Flag | Default | Effect |
|---|---|---|---|
| What can the server do? | DEBEZIUM_MODE | read-only | read-only exposes monitoring only; read-write adds lifecycle management; admin adds delete. Tools above the mode are never registered. |
| Which connectors are in scope? | DEBEZIUM_CONNECTOR_ALLOWLIST | (all) | When set, operations on other connectors are refused. |
| Which connectors are read-only forever? | DEBEZIUM_PROTECTED_CONNECTORS | (none) | Inspectable but never paused/reconfigured/deleted. |
| Can it delete? | DEBEZIUM_ALLOW_DELETE | false | delete_connector needs this and admin mode. |
| Preview without touching Connect | DEBEZIUM_DRY_RUN | false | Write/admin tools validate + log intent, then return. |
| Audit trail | DEBEZIUM_AUDIT_LOG | true | Emits a JSON line to stderr per guarded operation. |
| Interactive confirmation | (automatic) | — | Destructive & high-impact actions prompt the human to approve via MCP elicitation before running; clients without elicitation fall back to the *_ALLOW_* gate. |
| Credential redaction | (always on) | — | Secret-shaped config values are redacted before return. |
Tools
Read (read-only+): cluster_info, list_connectors, list_connector_plugins, get_connector, get_connector_config, get_connector_status, get_connector_topics
Write (read-write+): create_connector, update_connector_config, pause_connector, resume_connector, restart_connector, restart_task
Admin (admin): delete_connector (needs DEBEZIUM_ALLOW_DELETE)
Quickstart — add to your agent
Published on npm as @dockndevai/mcp-debezium. No clone or build needed — your MCP client runs it on demand with npx. Start in read-only mode; see .env.example for every variable and docs/CLIENTS.md for the full per-client guide.
Claude Code (CLI)
claude mcp add debezium -e CONNECT_URL="http://localhost:8083" -e DEBEZIUM_MODE="read-only" -- npx -y @dockndevai/mcp-debezium
Claude Desktop · Cursor · Windsurf — same block in claude_desktop_config.json, .cursor/mcp.json, or ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"debezium": {
"command": "npx",
"args": [
"-y",
"@dockndevai/mcp-debezium"
],
"env": {
"CONNECT_URL": "http://localhost:8083",
"DEBEZIUM_MODE": "read-only"
}
}
}
}
OpenAI Codex CLI — in ~/.codex/config.toml:
[mcp_servers.debezium]
command = "npx"
args = ["-y", "@dockndevai/mcp-debezium"]
env = { CONNECT_URL = "http://localhost:8083", DEBEZIUM_MODE = "read-only" }
VS Code (GitHub Copilot, Agent mode) — in .vscode/mcp.json:
{
"servers": {
"debezium": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@dockndevai/mcp-debezium"
],
"env": {
"CONNECT_URL": "http://localhost:8083",
"DEBEZIUM_MODE": "read-only"
}
}
}
}
Example prompts
- "Which connectors are FAILED, and why?"
- "Restart the failed task on
pg-orders." (needsread-write) - "Show the config for
mysql-inventory." (credentials come back redacted)
Run from source (development)
Prefer the published package above. To run from a clone:
npm install
npm run build
node dist/index.js # with the environment variables set
Develop
npm run dev
npm test # security policy + config redaction
npm run typecheck
Publishing
This server ships a server.json for the official MCP registry and an mcpName for npm ownership validation. See PUBLISHING.md for publishing to npm and listing on the MCP registry, Smithery, Glama, Cursor, and PulseMCP.
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.