Skillget

hookaido

mcp
SUMMARY

🪝 Durable webhook gateway with Caddyfile-style config. Receive → queue → deliver (pull, push, or exec). HMAC auth, retries, DLQ, hot reload.

README.md

Hookaido

CI
Release
Container
License
Latest Release

Webhook infrastructure that just works. Single binary. Zero runtime dependencies. Production-ready defaults.

Hookaido receives webhooks at the edge, queues them durably, and delivers them to your services — with retries, dead-letter queues, and cryptographic verification built in. Think of it as Caddy for webhooks: a compact config file, sensible defaults, and instant reloads.

Docs: https://nuetzliches.github.io/hookaido/

Why Hookaido?

Problem Hookaido
Webhooks hit your app directly — downtime means lost events Durable SQLite/WAL or PostgreSQL queue absorbs traffic; your services consume when ready
Retry logic scattered across services Exponential backoff, jitter, DLQ, and lease-based delivery — configured once
DMZ security headaches Pull mode by default: internal services fetch from the DMZ, no inbound holes
Complex multi-service deployment Single binary, one config file, go build and run
Webhook signature verification is error-prone Built-in HMAC-SHA256 verification with replay protection and secret rotation

Who Is It For?

  • SaaS integrators receiving webhooks from GitHub, Stripe, Shopify, or any HTTP callback provider
  • Platform engineers building DMZ-safe webhook ingestion without opening inbound firewall rules
  • DevOps teams triggering local deploy scripts, CI jobs, or automation from webhook events
  • Event-driven architectures that need a durable queue as lightweight middleware between producers and consumers

Key Features

Core

  • Durable queue — SQLite/WAL or PostgreSQL persistence with at-least-once delivery. In-memory mode for development.
  • Pull, push & exec — Pull API for DMZ-safe consumption, push dispatcher with concurrency control, subprocess execution for local scripts.
  • Hot reload — Change config, send SIGHUP or use --watch. No restarts for most changes.
  • Channel types — inbound (edge webhooks), outbound (API-published queues), internal (job queues).

Security

  • Ingress auth — HMAC signature verification with replay protection, Basic auth, forward auth callouts, rate limiting. Provider-compatible mode for GitHub and Gitea/Forgejo webhooks.
  • Outbound signing — HMAC-SHA256 on push delivery with multi-secret rotation windows and custom outbound headers.
  • Secret management — Env vars, files, Vault refs, and raw literals. Placeholder interpolation keeps secrets out of config.

Operations

  • Dead-letter queue — Failed messages land in the DLQ with full attempt history. Requeue or inspect via API.
  • Admin API — Health checks, queue inspection, backlog trends, publish/cancel/requeue operations.
  • Observability — Structured JSON logs, Prometheus metrics, OpenTelemetry tracing (OTLP).
  • MCP server — AI tooling integration with role-gated config inspection, queue diagnostics, and mutations.
  • VS Code extension — Syntax highlighting and snippets for Hookaidofile (editors/vscode/).

Quick Start

Download a binary from the latest release — no Go required.

Or build from source (requires Go 1.25+):

go build ./cmd/hookaido

Or use Docker (official GHCR image):

docker pull ghcr.io/nuetzliches/hookaido:latest
docker run -p 8080:8080 -p 9443:9443 \
  -e HOOKAIDO_PULL_TOKEN=mytoken \
  -v $(pwd)/Hookaidofile:/app/Hookaidofile:ro \
  -v hookaido-data:/app/.data \
  ghcr.io/nuetzliches/hookaido:latest

For immutable deployments, pin to a release tag (for example :v2.0.0) or digest.

Run locally:

export HOOKAIDO_PULL_TOKEN="mytoken"
./hookaido run --config Hookaidofile --db ./.data/hookaido.db

Ingress on :8080, Pull API on :9443, Admin on 127.0.0.1:2019 — ready to receive webhooks.

Configuration

A Hookaidofile is all you need. Caddy-inspired syntax, designed for readability:

Receive & pull (DMZ-safe):

ingress {
  listen :8080
}

pull_api {
  auth token env:HOOKAIDO_PULL_TOKEN
}

/webhooks/github {
  auth hmac env:HOOKAIDO_INGRESS_SECRET
  pull { path /pull/github }
}

Receive & push (direct delivery):

/webhooks/stripe {
  auth hmac env:STRIPE_SIGNING_SECRET
  deliver "https://billing.internal/stripe" {
    retry exponential max 8 base 2s cap 2m jitter 0.2
    timeout 10s
  }
}

Deliver via subprocess (no HTTP server needed):

/webhooks/github {
  auth hmac { provider github; secret env:GITHUB_SECRET }
  deliver exec "/opt/hooks/deploy.sh" {
    timeout 30s
    env DEPLOY_ENV production
  }
}

Placeholders keep secrets out of config: {$VAR}, {env.VAR}, {file./run/secrets/key}, {vars.NAME}.
Secret refs for auth/signing support env:, file:, vault:, and raw: schemes.

Full DSL reference and more examples: DESIGN.md | Recipes

Architecture

graph LR
    subgraph Providers
        GH[GitHub]
        ST[Stripe]
        WH[Any Webhook]
    end
    subgraph DMZ
        ING[Ingress :8080]
        Q[(SQLite / Postgres Queue)]
        ING --> Q
    end
    subgraph Internal Network
        SVC[Your Service]
    end
    GH & ST & WH -->|POST| ING
    SVC -->|pull /pull/…| Q
    Q -->|push deliver| SVC

Default: dmz-queue pull — Hookaido sits in the DMZ. Internal workers pull over HTTPS. No inbound firewall rules needed.

Admin API

GET /healthz?details=1 — queue diagnostics, backlog trends, operator action playbooks.

POST /messages/publish — inject messages programmatically.

GET /dlq / POST /dlq/requeue — inspect and recover dead-lettered messages.

GET /backlog/trends — time-series backlog analysis with derived signals.

Full endpoint list → DESIGN.md

Observability

observability {
  access_log  { output stderr; format json }
  runtime_log { level info; output stderr; format json }
  metrics     { listen ":9900"; prefix "/metrics" }
  tracing     { collector "https://otel.example.com/v1/traces" }
}

Prometheus counters for ingress, adaptive backpressure (reason-labeled), pull (dequeue/ack/nack/conflicts/leases), and push delivery, plus queue depth gauges and SQLite store contention metrics (write/dequeue/checkpoint durations, busy/retry, tx commit/rollback). OpenTelemetry traces with full OTLP/HTTP configuration.

MCP Server (AI Integration)

./hookaido mcp serve --config Hookaidofile --db ./.data/hookaido.db --role read

Exposes queue state, config inspection, health diagnostics, and backlog analysis as structured tools for AI assistants. Role-gated access: read, operate, admin.

Release Verification

Releases ship with signed checksums (Ed25519), SPDX SBOM, and GitHub provenance attestations:

./hookaido verify-release --checksums hookaido_v2.0.0_checksums.txt \
  --public-key release-signing-key.pub \
  --require-provenance

Requirements

  • Pre-built binaries: no dependencies (download from Releases)
  • Build from source: Go 1.25+
  • Docker: use the official image ghcr.io/nuetzliches/hookaido (or build locally), see Docker quickstart
  • No external runtime dependencies

Documentation

Document Purpose
docs/ User-facing documentation (getting started, configuration, APIs, deployment)
DESIGN.md Canonical DSL and API specification
CONTRIBUTING.md Contribution workflow, development setup, and maintainer notes
CHANGELOG.md User-visible changes per release
SECURITY.md Vulnerability reporting and security response policy
SUPPORT.md Support channels and issue quality guidance

License

Apache-2.0 — see LICENSE.

Yorumlar (0)

Sonuc bulunamadi