notifyd

บริการแจ้งเตือนแบบโฮสต์เอง (อีเมล, SMS, push, in-app) ในไบนารี Rust เดียวบน Postgres; MCP server ของมันช่วยให้เอเจนต์จัดการอินสแตนซ์ได้: สรุปย่อ, งาน, การลองใหม่, การระงับ, การส่งทดสอบ

GitHub

ลองใช้ MCP นี้ผู้สนับสนุน

เอกสาร

notifyd

Send email, SMS, WhatsApp, push and in-app notifications from one API call.
One 12 MB Rust binary, PostgreSQL only. Queues, retries, provider failover and quiet hours are handled for you, and an AI agent can run it over MCP.

Quick Start • Clients • Examples • Agent operations • API Reference • Architecture • Benchmarks • llms.txt • Contributing

What it is

Your app has to tell people things: a password reset, a shipped parcel, a newsletter, a red badge in the corner. notifyd is the small server that does all of it, so your code makes one call and never has to think about providers, rate limits, retries or time zones again.

                 ┌───────────────────────────── notifyd ──────────────────────────────┐
your app ─────▶  │ POST /v1/send ─▶ queue ─▶ priority ─▶ pacing ─▶ retry ─▶ failover  │ ─▶ email · sms · whatsapp
                 │                                                                    │    push · in-app · telegram · slack · discord
your agent ───▶  │ POST /mcp ─▶ digest · jobs · retry · suppressions · settings       │
                 └───────────────────────── PostgreSQL only ──────────────────────────┘

Small and fast. One 12 MB binary, a 44 MB image, 13 MB of RAM idle. It accepts 44 000 notifications per second and drains 3 500 per second on a laptop (method). No Redis, no message broker, no dashboard to host: PostgreSQL is the only dependency.
Nothing gets lost. A password reset always goes before a campaign. When a provider says "slow down", that channel pauses for exactly the time asked and resumes in priority order; when it fails, a second provider takes over. Retries, idempotency and quiet hours in each recipient's time zone are built in.
Your providers, your data. Resend, Cloudflare Email, any SMTP, AgentMail, Telnyx, Twilio, APNs, Web Push, FCM. Self-hosted, MIT.
Operated by an API or an AI agent. No admin UI: a digest endpoint says what needs attention and what to do, and the same operations are MCP tools, so the person on call can be an agent.

Three instances run in production today, one per company, operated this way.

_{60 s explainer, invented data. MP4 · Remotion source}

Send in one call

curl -X POST https://notifyd.example.com/v1/send \
  -H "X-Api-Key: sk_myapp_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "channels": ["email", "in_app"],
    "subscriber_id": "user-1",
    "subject": "Your report is ready",
    "body": "Hey {{first_name}}, the analysis you requested is complete.",
    "vars": {"first_name": "Alice"},
    "priority": "normal",
    "idempotency_key": "report-42-ready"
  }'

Flat REST, curl works; TypeScript and Python clients exist. Retries are safe (idempotency_key), scheduling is a field (scheduled_at), a marketing campaign is POST /v1/batch with thousands of subscribers per call and it lands in the bulk lane so it never delays a password reset. Follow any send with GET /v1/jobs/:id. Runnable examples in every language: examples/.

Run it from a terminal

The same binary is the operator's CLI, against any instance (the server host has ADMIN_API_KEY in its environment, so it just works there):

notifyd digest                       # what needs attention, with the action for each finding
notifyd jobs --status failed         # --project, --channel, --topic, --recipient, --since 24h, --limit, --json
notifyd job <id>                     # provider, attempts, delivery events
notifyd retry <id>                   # re-queue after fixing the cause
notifyd send-test --project myapp --channel email --to you@example.com
NOTIFYD_URL=https://notifyd.example.com NOTIFYD_ADMIN_API_KEY=… notifyd digest   # from your laptop

Let your agent run it

Most notification tools were designed for a human clicking through a dashboard. notifyd exposes the operator's job as tools, with the detail a human operator would need:

1. One call says what needs attention. GET /v1/admin/digest ranks findings and tells you the action for each one, in JSON or Markdown:

# notifyd digest — last 1d
Instance: commit e14e6f3, up 3d, email resend (+ smtp fallback), sms telnyx

## Findings
- **warning** — Primary email provider `resend` is resting for 47s after refusing messages; `smtp` is delivering.
  _Nothing lost. Check the primary provider's status page; if it repeats, lower EMAIL_RATE_PER_SEC or move the primary role to the other provider._
- **warning** — Bounce rate 5.3 % over the window (14 bounced / 263 delivered).
  _Above 5 % providers throttle or suspend the sender. Clean the recipient list; suppressions are applied automatically._
- **warning** — 3 job(s) failed in the last 1d (0.1 % of terminal jobs). Top cause: 422 unverified sender domain.
  _Inspect with list_jobs(status=failed); permanent errors need a fix on the caller side, then retry_job._

## Queue          pending 0, retry 2, processing 0
## Outcomes       email/resend 4 812 sent, 3 failed · in_app 1 203 sent
## Latency        email p50 0.6s, p95 2.1s (scheduled → accepted by provider)
## Deliverability delivered 4 790, bounced 14, complained 0, unsubscribed 9

2. The same operations as MCP tools. POST /mcp is a Streamable HTTP MCP server (current spec revision, legacy initialize kept). Add it to Claude Code, Claude Desktop, Cursor or your own agent:

{ "mcpServers": { "notifyd": {
  "type": "http", "url": "https://notifyd.example.com/mcp",
  "headers": { "Authorization": "Bearer ${NOTIFYD_ADMIN_API_KEY}" } } } }

Tool	What the agent can do
`digest`	Ranked findings with actions, queue, outcomes, latency, deliverability, per project
`list_jobs`, `get_job`	Filter by project, channel, status, recipient, time; see provider, attempts, last error
`retry_job`, `cancel_job`	Act on a stuck or wrong send
`list_projects`, `update_project`	Sender identity (`from_email`, `from_name`), channels, inbound rate limit, bulk `send_window` in the recipients' timezone
`list_suppressions`, `add_suppression`, `release_suppression`	Suppression list with `all` or `marketing` scope
`template_metrics`	Sent, failed, bounced, opened per template
`send_test`	Prove the pipeline end to end on any channel

Every tool carries readOnlyHint / destructiveHint annotations and an outputSchema. A read-only operator key (READONLY_API_KEY) exposes only the read tools, for an agent that reports but must not act. Every MCP call is audited.

3. The digest comes to you. DIGEST_NOTIFY=telegram:<chat id> (or a Slack / Discord webhook) and the findings above land in your chat when something needs attention; notifyd digest --to telegram:… sends one now.

4. Everything an agent needs to integrate is in the repo. docs/llms.txt is the whole API in plain text for a context window; three Agent Skills ship in skills/ (notifyd-operate, notifyd-integrate, notifyd-deploy):

npx skills add rmzlb/notifyd

Published on the official MCP registry as mcp-name: io.github.rmzlb/notifyd (server.json). Full operator guide: docs/AGENT.md.

Features

Channels

Email — Resend, Cloudflare Email Service, AgentMail, any SMTP (lettre), attachments, per-project sender identity
SMS — Telnyx or Twilio, swap with one variable
WhatsApp — Telnyx
Push — APNs natively (.p8 token auth, HTTP/2, badge, sound, thread id, silent pushes, dead tokens dropped), Web Push (VAPID), FCM
In-app inbox — REST + realtime SSE (EventSource), read / archive / star, unread badge, multi-replica through Postgres NOTIFY

Delivery engine

Priorities — critical, normal, bulk lanes; /v1/batch and campaign tags land in bulk
Pacing — token bucket per channel (EMAIL_RATE_PER_SEC); a provider 429 pauses that channel for Retry-After without consuming an attempt, other channels keep flowing, and when it resumes the claim order puts critical first
Retries — 30 s → 2 m → 10 m → 30 m → 2 h with jitter, 4xx fail fast, rejected batches fall back item by item
Failover — second email provider with a circuit breaker (EMAIL_FALLBACK_PROVIDER)
Send windows — quiet hours per project, evaluated in each subscriber's timezone
Segments — batch to "plan = pro and country in FR, BE" with a small filter vocabulary compiled to bound SQL; preview the count first
Scheduling, idempotency, stuck-job reaper, batch idempotency

Governance

Unsubscribe — RFC 8058 List-Unsubscribe one-click on every marketing email, suppression scopes all / marketing
Topics and preferences — a topic on any send ("tips", "billing"); subscribers opt out per topic and channel, honoured at enqueue; the unsubscribe page offers "this topic only" before "all marketing"
Multi-project — one instance, many projects, isolated by API key, key rotation with a grace period
PII masking in logs, audit log of every mutation, per-project rate limit

Operations

Digest, MCP server, Agent Skills, llms.txt
Metrics — /v1/metrics, /v1/metrics/prometheus, per-template metrics
Open and click tracking — own pixel and signed redirect links for every email provider, on marketing email by default (transactional links stay untouched), opened_at / clicked_at on the job, off per project or per request
Webhooks — delivery events to your endpoints
Workflows — event-triggered multi-step sequences, state in Postgres
Templates — {{variable}} substitution, stored per project

Quick Start

Docker (recommended)

The image reads its configuration from environment variables; no config file to mount.

git clone https://github.com/rmzlb/notifyd.git && cd notifyd
cat > .env <<'EOF'
JWT_SECRET=change-me-32-random-chars-minimum
ADMIN_API_KEY=change-me-32-random-chars-minimum
RESEND_API_KEY=re_xxx
EMAIL_FROM=notifications@yourdomain.com
EOF
docker compose up -d          # notifyd + Postgres 16, http://localhost:3400

Create a project and get its API key:

curl -s -X POST http://localhost:3400/v1/admin/projects \
  -H "X-Api-Key: $ADMIN_API_KEY" -H "Content-Type: application/json" \
  -d '{"id":"myapp","name":"My app","channels":["email","in_app"],"from_email":"hello@yourdomain.com"}'
# → {"project": {"id": "myapp", "api_key": "sk_myapp_…", …}}

Prebuilt image, linux/amd64 and linux/arm64: ghcr.io/rmzlb/notifyd.

Binary, crate, Nix or source

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/rmzlb/notifyd/releases/latest/download/notifyd-installer.sh | sh
brew install rmzlb/tap/notifyd             # macOS and Linux, Homebrew
cargo binstall notifyd                     # prebuilt from the GitHub release
cargo install notifyd                      # build from crates.io
nix run github:rmzlb/notifyd               # flake: packages, devShell, NixOS module
# then
DATABASE_URL=postgres://notifyd:pass@localhost:5432/notifyd \
JWT_SECRET=… ADMIN_API_KEY=… RESEND_API_KEY=… EMAIL_FROM=… notifyd

Release binaries: Linux x86_64 and aarch64, macOS Intel and Apple Silicon. NixOS: services.notifyd.enable = true; with an environmentFile (see flake.nix).

No provider yet? EMAIL_PROVIDER=log prints emails instead of sending them.

Verify

curl http://localhost:3400/v1/health
# → {"status":"ok","db":"ok","version":"0.2.2","commit":"…","uptime_seconds":12}

→ Full setup, TOML alternative, production notes: docs/SETUP.md

API at a glance

Project endpoints take X-Api-Key: sk_<project>_…; operator endpoints take the admin (or read-only) key, as X-Api-Key or Authorization: Bearer. Inbox endpoints also accept a subscriber JWT.

Method	Endpoint	What it does
`POST`	`/v1/send`	Send on one or several channels
`POST`	`/v1/batch`	Send to a list of subscribers or to a `segment` filter (bulk lane, idempotent)
`POST`	`/v1/segments/preview`	Count and sample the subscribers a segment matches
`GET`	`/v1/jobs/:id`	Status, provider, attempts, last error
`GET`	`/v1/inbox/:id` · `/stream`	In-app inbox, SSE realtime stream
`POST`	`/v1/workflows/trigger`	Trigger an event-based workflow
`GET`	`/v1/admin/digest`	What needs attention, with actions
`GET`	`/v1/admin/jobs` · `/:id` · `POST …/:id/retry` · `/admin/send-test`	Operator view and actions (also the CLI's transport)
`PATCH`	`/v1/admin/projects/:id`	Sender, channels, rate limit, send window
`POST`	`/mcp`	MCP server (Streamable HTTP)
`GET`	`/v1/metrics/prometheus`	Prometheus exposition
`GET`	`/u/:token`	One-click unsubscribe landing

→ Every endpoint with examples: docs/API.md, or feed docs/llms.txt to your agent.

vs. the alternatives

	Novu	Knock / Courier / SuprSend	notifyd
Infra	MongoDB + Redis + 4 app containers	Hosted SaaS	Postgres only, one 44 MB image
Setup	30+ min	Signup + dashboard	`docker compose up` (2 min)
Language	Node.js (multiple services)	N/A (hosted)	Rust (single binary)
Memory at idle	1.1 GB across 6 containers (measured)	N/A	13 MB, 23 MB while draining 100k jobs (method)
Images to pull	1.4 GB	N/A	44 MB
Throughput	—	quota-bound	44k jobs/s enqueued, 3.5k jobs/s drained (benchmarks)
Provider 429	job fails	managed	channel paused for `Retry-After`, attempt not consumed, failover provider tried first
Priorities / send windows	❌	✅	✅ critical → bulk lanes, per-subscriber timezone windows
Ops surface	React dashboard	dashboard + API	digest endpoint, MCP server, Agent Skills, Prometheus
Realtime	WebSocket	WebSocket	SSE, native `EventSource`, multi-replica
Self-hosted	✅ (heavy)	❌	✅ one container per company
Cost	Free tier / paid	per notification	Free forever, MIT

Every number in this table was measured by us; the Novu figures come from its own community docker-compose, idle, on the same machine and with the same tool as ours. Method, hardware and bias disclaimer in docs/BENCHMARKS.md.

Clients

Both clients cover the whole API (send, batch, jobs, subscribers, preferences, templates, workflows, suppressions, inbox) and raise a typed error on any non-2xx answer.

TypeScript / JavaScript — npm i notifyd-sdk (Node 18+, browsers, edge runtimes; zero dependencies)

import { createNotifydClient } from 'notifyd-sdk';

const notifyd = createNotifydClient({ url: process.env.NOTIFYD_URL!, apiKey: process.env.NOTIFYD_API_KEY! });

const { jobIds } = await notifyd.send({
  channels: ['email', 'in_app'], subscriberId: 'user-42',
  subject: 'Your order shipped', body: 'Hi {{first_name}}, parcel {{parcel}} is on its way.',
  vars: { first_name: 'Alice', parcel: 'FR-2041' }, idempotencyKey: 'order-2041-shipped',
});
const job = await notifyd.getJob(jobIds[0]);   // status, provider, attempts, delivery events

// Browser inbox: subscriber token from your backend, live updates over EventSource
const inbox = createNotifydClient({ url, subscriberToken });
const stream = await inbox.openInboxStream('user-42', { onMessage: (e) => {
  const event = JSON.parse(e.data);
  if (event.type === 'new_notification') showToast(event.notification);
  if (event.type === 'count_update') updateBadge(event.unread_count);
} });

Python — pip install notifyd-sdk (3.9+, sync and asyncio, httpx only) — clients/python

from notifyd import Notifyd

nd = Notifyd(os.environ["NOTIFYD_URL"], api_key=os.environ["NOTIFYD_API_KEY"])
result = nd.send(channels=["email", "in_app"], subscriber_id="user-42",
                 subject="Your order shipped", body="Hi {{first_name}}, parcel {{parcel}} is on its way.",
                 vars={"first_name": "Alice", "parcel": "FR-2041"}, idempotency_key="order-2041-shipped")
print(nd.get_job(result["job_ids"][0])["status"])

Any other language: the API is flat JSON over HTTP and docs/llms.txt is the whole contract on one page.

Workflows

Multi-step sequences triggered by an event, state in Postgres, survive restarts. Steps run in order; a condition jumps to a step index.

curl -X POST http://localhost:3400/v1/workflows -H "X-Api-Key: sk_myapp_xxx" -d '{
  "id": "welcome-series", "name": "Welcome series", "trigger_event": "user.signup",
  "steps": [
    {"type": "send", "channel": "email", "template": "welcome"},
    {"type": "delay", "duration_secs": 86400},
    {"type": "condition", "field": "payload.plan", "operator": "eq", "value": "pro", "on_true": 4},
    {"type": "send", "channel": "email", "template": "nudge"}
  ]}'

curl -X POST http://localhost:3400/v1/workflows/trigger -H "X-Api-Key: sk_myapp_xxx" \
  -d '{"event": "user.signup", "subscriber_id": "user-42", "payload": {"plan": "free"}}'

Step types: send, delay, condition (inbox.is_read or payload.<key>) and digest (collect events for a while, then send one message). Full script: examples/welcome-series.sh.

Configuration

Environment variables are the primary interface (that is what the image and the compose file use); a notifyd.toml is accepted for local development. Required: DATABASE_URL, JWT_SECRET, ADMIN_API_KEY. Then one provider:

Variable	Purpose
`EMAIL_PROVIDER`	`resend` (default when `RESEND_API_KEY` is set), `cloudflare`, `smtp`, `agentmail`, `log`
`EMAIL_FROM`, `EMAIL_FROM_NAME`	Instance default sender; projects can override
`EMAIL_FALLBACK_PROVIDER`	Second provider on 429 / 5xx
`EMAIL_RATE_PER_SEC`	Outbound pacing per replica
`SMS_PROVIDER`, `SMS_FROM`	`telnyx` or `twilio` with their credentials
`PUBLIC_URL`	Base URL for one-click unsubscribe links
`READONLY_API_KEY`	Optional read-only operator key

→ Every variable, per provider: docs/CONNECTORS.md and docker-compose.yml. TOML reference: notifyd.toml.example.

Architecture

                 ┌──────────────────────── notifyd (one binary) ───────────────────────┐
 HTTP /v1, /mcp ─▶ axum API ─▶ jobs table ─▶ worker: claim (SKIP LOCKED, by priority) │
                 │                              ├─ pacer per channel, channel pause on 429│
                 │                              ├─ connectors (email/sms/whatsapp/push/in-app)
                 │                              ├─ failover breaker, retries, reaper      │
                 │                              └─ webhooks, metrics, audit               │
                 │  SSE hub ◀── Postgres NOTIFY ── (any replica)                          │
                 └───────────────────────────────┬──────────────────────────────────────┘
                                                 ▼
                                           PostgreSQL 16

src/
├── api/              # routes: send, batch, jobs, inbox, subscribers, templates, workflows, webhooks, admin ops, health
├── connectors/       # email (resend, cloudflare, smtp, agentmail, log), sms, whatsapp, push, in_app
├── worker.rs         # claim by priority, batch context, finalize, retries, failover
├── pacing.rs         # token buckets and channel pauses
├── failover.rs       # provider circuit breaker
├── ops.rs            # digest, findings, operator actions
├── mcp.rs            # MCP server (tools, annotations, audit)
├── send_window.rs    # quiet hours in the subscriber's timezone
├── unsubscribe.rs    # List-Unsubscribe tokens and landing
├── sse.rs            # inbox stream, Postgres NOTIFY fan-out
├── workflow_engine.rs, templates.rs, webhooks.rs, deliverability.rs, metrics.rs, pii.rs, middleware.rs
migrations/           # SQL, applied at start-up
skills/               # Agent Skills: operate, integrate, deploy
server.json           # MCP registry entry
flake.nix             # Nix package, devShell, NixOS module
dist-workspace.toml   # cargo-dist: release binaries and installer

~14 000 lines of Rust, no unsafe. 12 MB binary (16 MB static musl in the image), 44 MB image, 13 MB RSS idle, 23 MB while draining 100 000 jobs. → docs/ARCHITECTURE.md

Status

notifyd is a 0.x running in production for three companies. Not done yet: Swift and Kotlin packages; no dashboard, A/B testing or inbound email by design. Order and sizes in docs/ROADMAP.md. Breaking changes are announced in release notes; the queue schema is migrated automatically.

Documentation


📦 Setup	Local dev, Docker, production
🧪 Examples	curl, TypeScript, Python, a 10k campaign, a workflow, a browser inbox, MCP config
🐍 Python client · TypeScript client	Full-API clients, typed errors, contract tests
🔌 API reference	Every endpoint with curl / TypeScript / Rust examples
🤝 Agent operations	Digest, MCP tools, read-only key, how an agent runs an instance
🔌 Connectors	Providers, environment variables, adding one
🏗️ Architecture	Queue, priorities, pacing, SSE, connectors
📈 Benchmarks	Footprint, throughput, how to reproduce
🚀 Deployments	One instance per company, runbook
📝 Writing	A notification queue on PostgreSQL alone: what `SKIP LOCKED` does not give you
🗺️ Roadmap	What comes next, in order, with sizes; what is deliberately not planned
📣 Visibility	Registries and launch channels
🤖 llms.txt	The API in plain text for agents

Contributing

Issues and pull requests are welcome; good first issue is the place to start, and big features begin with an issue. Read CONTRIBUTING.md.

git clone https://github.com/YOUR_USERNAME/notifyd.git && cd notifyd
cargo test && EMAIL_PROVIDER=log DATABASE_URL=… JWT_SECRET=dev ADMIN_API_KEY=dev cargo run

License

MIT.

Built with 🦀 in Grenoble, France 🏔️