Signal Pipeline

Ingest signals from any source via MCP server, REST API, webhooks, or the Connector SDK. Normalize everything into a unified schema for AI analysis.

Flexible Ingestion Methods

Connect any signal source using the method that fits your infrastructure.

MCP Server

Built-in MCP server for direct AI client connectivity. Connect from Claude Desktop or any MCP-compatible client.

REST API

Full REST API for programmatic ingestion, automation scripts, and CI/CD pipelines.

Webhooks

Receive push-based signals in real time. Supports HMAC verification, retry logic, and batched payloads.

Connector SDK

Build custom connectors in TypeScript for any source. Full type safety with hot-reload.

Unified Normalization

Every signal is transformed into a consistent schema so AI models can analyze across sources.

1

Unified Schema — Raw payloads are mapped to a standard envelope with source, timestamp, severity hint, and payload fields.

2

Metadata Extraction — Stack traces, CVE identifiers, affected endpoints, and service names are automatically parsed.

3

Schema Validation — Malformed signals are quarantined with detailed error reports for debugging.

Deduplication & Grouping

Duplicate signals are automatically detected and grouped. A single Sentry error firing 200 times becomes one enriched signal, not 200 tickets.

  • Content-hash based dedup
  • Configurable time windows
  • Cross-source grouping

Real-time Processing

Signals flow through the pipeline in seconds, not minutes. Stream processing ensures your team is alerted before issues escalate.

  • Sub-second ingestion
  • Backpressure handling
  • Pipeline health dashboard

For Managers

The pipeline health dashboard shows you at a glance whether signals are flowing from all connected tools. You will know immediately if a connector goes silent — before your team does.

Get Early Access to Signal Pipeline

Join the waitlist to be first in line. Connect webhooks, configure polling, or build a custom connector on day one.