Getting Started FAQ#

Which mode should I choose first?#

Use this default decision path:

streamable-http: best fit for modern MCP transport in production
sse: broad compatibility with existing MCP clients and integrations

If you are unsure, start with sse, complete validation, then migrate to streamable-http.

Why do I get `401 unauthorized`?#

Check the following in order:

MCP_AUTH_ENABLED=true is actually set in runtime environment.
MCP_AUTH_MODE is one of apikey, bearer, basic.
For apikey mode, MCP_AUTH_API_KEY is non-empty and matches request value.
API key is sent via X-Api-Key header or api_key query parameter.
Container/process was restarted after configuration changes.

Is API key header or query parameter preferred?#

Both are supported, but use request headers in production whenever possible:

1
2
3
curl -sS -N \
  -H "X-Api-Key: ChangeMe-Strong-Key-123!" \
  http://127.0.0.1:8080/api/aggregate/sse

Use query parameter mainly for quick local checks.

How can I reduce response size for AI agents?#

Use summary-focused tools before requesting full payloads.
Use pagination-capable tools for large result sets.
Disable unnecessary services via MCP_DISABLED_SERVICES.
Apply request rate limits to avoid burst overload.

Can I enable only a subset of services?#

Yes. You can explicitly enable or disable services:

1
2
export MCP_ENABLED_SERVICES="kubernetes,helm,prometheus"
export MCP_DISABLED_SERVICES="kibana,jaeger"

Keep configuration consistent to avoid overlapping expectations in team environments.

What is a safe production baseline?#

Enable authentication and rotate secrets regularly.
Protect upstream credentials (Grafana, Prometheus, Kibana, etc.).
Enable structured logging and audit if required.
Monitor /health and key service integrations continuously.
Run load tests against realistic tool-call patterns.

Where should I go next?#