Skip to Content
Configuration

Configuration

Each project in GlassFlow AI Runtime has its own pipeline configuration. You can manage it through the web UI or the control plane API.

Pipeline

The pipeline processes incoming OTLP data before forwarding it to an AI agent. It has three parts:

Filter

An optional expression-based filter that drops records that don’t match. Uses the expr  language.

Example expressions:

severityNumber >= 17

Only forward ERROR and FATAL logs.

resourceAttributes["service.name"] == "my-service"

Only forward logs from a specific service.

severityNumber >= 9 && resourceAttributes["k8s.namespace.name"] == "production"

Only forward warnings and above from production.

Transform

Optional field transformations applied to each record after filtering. Each transform extracts or computes a value and adds it to the record.

FieldDescription
expressionAn expr expression evaluated against the record
output_nameThe field name to write the result to
output_typeThe result type: string, int, int64, float64, bool

Example:

ExpressionOutput nameType
resourceAttributes["service.name"]servicestring
severityNumber >= 17is_errorbool

Agent endpoint

The URL where the pipeline sends batched, filtered records. Set this to your AI agent’s processing endpoint (e.g. http://my-agent:8000/process).

The pipeline only activates for projects that have an agent endpoint configured. Leave it empty to disable pipeline processing.

The pipeline batches up to 10 records per request. Each batch is sent as a JSON array in the POST body, with X-Project-ID and X-Request-ID headers.

Sinks

Sinks receive enriched output from AI agents (via the NATS output stream) and dispatch it to external destinations.

Webhook

Sends the agent output as an HTTP request:

FieldDescription
URLThe webhook endpoint
MethodHTTP method (POST, PUT, etc.)
HeadersCustom headers (JSON object)

Slack

Posts the agent output to a Slack channel via an incoming webhook:

FieldDescription
Webhook URLSlack incoming webhook URL

API keys

Each project has API keys used to authenticate:

  • OTLP ingest — your applications include the key in the X-API-Key header when sending telemetry
  • Agent output — the GlassFlow SDK uses the key to send enriched results back to the receiver

Create and manage keys in the project’s API Keys page.

API reference

All configuration is available via the control plane REST API:

EndpointMethodDescription
/api/v1/projects/{id}/pipelineGET / PUTPipeline config (filter, transforms)
/api/v1/projects/{id}/bridgeGET / PUTAgent endpoint URL
/api/v1/projects/{id}/sinksGET / PUTSink configuration
/api/v1/projects/{id}/keysGET / POSTAPI key management