Skip to main content
Streamkap provides tools that enable AI agents to interact with your CDC infrastructure and build on top of it. Agents can monitor pipeline health, troubleshoot issues, manage resources, and automate operational tasks — through natural language, CLI, or programmatic interfaces.

​
What Agents Can Do

Connect an agent to Streamkap and ask it to:
  • “Give me an overview of my infrastructure and flag anything unhealthy”
  • “Check the logs for errors in the last hour and diagnose the root cause”
  • “Create a new pipeline from my PostgreSQL source to Snowflake”
  • “Show me consumer group lag for all my pipelines”
  • “Pause all sources tagged as ‘staging’”
  • “What schemas are available in my schema registry?”
  • “Export my billing usage for the last month”
Agents have access to the full Streamkap API surface — sources, destinations, pipelines, transforms, topics, schemas, logs, metrics, alerts, and more.

​
Integration Paths

IntegrationBest ForHow It Works
MCP ServerAI agents in IDEs and chat interfacesConnects to Claude, Cursor, Windsurf, VS Code Copilot, and other MCP-compatible clients. Agents call Streamkap tools through natural language.
CLIScripts, CI/CD, and agent pipelinesOutputs JSON when piped and skips confirmations in non-interactive mode — designed for agents to invoke directly.
REST APICustom integrations and direct API callsFull programmatic access to all Streamkap operations. Use when building custom tooling or integrating with platforms that don’t support MCP.
TerraformInfrastructure as codeDeclare your Streamkap resources in HCL. Agents can generate, review, and apply Terraform configurations. Also works with Terraform MCP servers for agent-driven infrastructure management.

​
Credentials

All integration paths require authentication. You have two options:
  • API Tokens — standalone Client ID and Secret for API access
  • Project Keys (recommended for agents) — unified credential files that bundle API access, Kafka access, and MCP tool scoping. Project Keys support tool scoping — you can restrict which MCP tools an agent can call without changing your server configuration.

​
MCP Server

The MCP Server is the primary way to connect AI agents to Streamkap. It exposes tools covering every Streamkap API operation.

​
Setup

The MCP server runs in two modes — remote (hosted) or local (via npx) — with setup guides for all major AI clients. See MCP Server for detailed instructions.

​
CLI for Agents

The CLI is designed to work in agentic pipelines alongside interactive use:
  • Auto-JSON output — detects when output is piped (non-TTY) and switches to JSON automatically
  • No confirmation prompts — destructive commands execute without confirmation in non-interactive mode
  • Quiet mode — --quiet suppresses all non-data output for clean parsing
# Auto-JSON when piped — no flag needed
streamkap sources list | jq '[.result[] | select(.connector_status == "Broken") | .name]'

​
Getting Started

  1. Create credentials — either an API Token or a Project Key (recommended for agents — supports tool scoping)
  2. Pick your integration — MCP Server for AI agents, CLI for scripts, REST API for custom tooling
  3. Connect and go — setup is quick for any integration path