CLI v2 with MCP, Request issues, Discord notifications - Hookdeck March 2026

Issues now catch request-level problems, Discord gets real-time Event Gateway alerts, the CLI gets an MCP server, and Static IPs are self-serve. That's just some of what shipped in March.

CLI v2: MCP server for AI agents

The Hookdeck CLI is no longer a tool just for developers, it's also an integration point for AI agents with the addition of a built-in MCP server.

Start it via hookdeck gateway mcp and configure it to run with your MCP client (Claude Desktop, Cursor, etc.):

{
  "mcpServers": {
    "hookdeck": {
      "command": "hookdeck",
      "args": ["gateway", "mcp"]
    }
  }
}

The server exposes 11 tools: projects, connections, sources, destinations, transformations, requests, events, attempts, issues, metrics, and help. These are mostly read and inspect tools — the things an agent needs to answer "what's failing and why?" without opening the Dashboard. The connections tool also supports pause and unpause so an agent can act on what it finds.

What this looks like in practice

We point Claude at a production Hookdeck project and ask it to analyze 30 days of data broken down by connection, looking for failure rates and problems. Using the Hookdeck MCP tools, it pulls connections and metrics, builds a failure dashboard, and flags a Customer.io to PostHog connection where events are failing with HTTP 400s because distinct_id is null.

From there it digs into the specific events, identifies that the transformation doesn't handle cases where Customer.io's identifier fields are missing, and generates an updated transformation with a fallback chain so it's ready to copy and paste into the Hookdeck Event Gateway dashboard. No scripting. No manual log trawling. Just a conversational investigation.

v2 also adds hookdeck gateway issue and hookdeck gateway metrics command groups, a --local flag for project-scoped credentials, opt-out telemetry, and a breaking rename of --config to --hookdeck-config.

See the full CLI v2 release post →

Request problems in Issues

Issues in Event Gateway now surface request problems alongside existing delivery, transformation, and backpressure types.

Request issues trigger when incoming webhooks are rejected by a source, such as for authentication failures, malformed payloads, and other ingestion problems that could go unnoticed. Now you can view the rejected request's metadata and HTTP details, plus bulk retry rejected requests.

Discord notifications

Discord joins Slack and Microsoft Teams as a notifications channel in Event Gateway. Here's what you can do:

• New Issue Alerts: Get notified instantly when new event issues arise
• Issue Updates: Track when issues are being acked or ignored
• Resolution Notifications: Know immediately when issues are resolved
• Flexible Channel Support: Connect to any text or announcement channel in your Discord server

Configure Discord notifications →

Self-serve Static IP

Static IP is now a self-serve add-on. Enable or disable it from the Delivery section in your project settings. Static IP ensures webhook deliveries arrive from a fixed set of IP addresses, so you can configure firewall rules and IP allowlists on your endpoints without coordination overhead.

Outpost releases

Outpost v0.15.0 added a metrics API with Postgres and ClickHouse backends, while the portal got UI improvement. v0.14.0 focused on infrastructure reliability including scheduler monitor improvements and native SDK StreamingPull for GCP Pub/Sub.

New source types

Four new webhook source types for Event Gateway: Uber, ShipHero, Chaindots, and Faundit. Each includes HMAC signature verification. These came directly from user requests — if you need a source type we don't support yet, let us know.

60x faster payload search

Searching webhook payloads is one of the first things you do when something breaks, so it needs to be fast. We wrote up how we brought average latency down to ~400ms (p99 under 1.4s) with Hash-bucketed storage and Variable-window scanning. The result is search that feels instant.

Read the full technical deep-dive →

AI integration guides

We published two new guides for connecting webhooks to AI agent workflows:

• GitHub automation with Hookdeck + Trigger.dev + Claude — Learn how to automate PR code reviews, issue labeling, and deployment summaries to Slack.

• Claude Code Channels + Hookdeck CLI — Push external events directly into running Claude Code sessions via webhook channels and Hookdeck CLI.

Community: hookdeck-deploy-cli

hookdeck-deploy-cli is a community-built CLI from the kind folks at Toppy that brings declarative, infrastructure-as-code workflows to Hookdeck — without needing Terraform. Define your sources, destinations, connections, and transformations in hookdeck.jsonc manifest files, then deploy with a single command.

It supports multi-environment overrides, automatic dependency ordering, drift detection, and variable interpolation. If you want version-controlled Hookdeck configuration that deploys from CI, this is worth a look.

npm install -D @toppy/hookdeck-deploy-cli

Check it out on GitHub →