nn-webhooks-sdk (Python)

Official Python SDK for NimbusNexus Webhooks — publish events, manage your endpoints and deliveries, and verify the webhooks you receive.

pip install nn-webhooks-sdk

Verify an incoming webhook (subscribers)

When webhookd delivers a webhook it signs the body with your endpoint's signing secret. Always verify the signature before trusting the payload — it proves the request really came from webhookd and wasn't tampered with or replayed.

from nn_webhooks import verify

# In your webhook handler — pass the RAW request body bytes (do not re-serialize the JSON):
ok = verify(
    secret=ENDPOINT_SIGNING_SECRET,
    raw_body=request.body,
    signature=request.headers["X-Webhook-Signature"],
    timestamp=request.headers["X-Webhook-Timestamp"],
)
if not ok:
    return Response(status_code=400)  # forged, tampered, or outside the 300s replay window

Publish an event (producers)

from nn_webhooks import Client, WebhooksAPIError

with Client("https://webhooks.example.com", api_key="eyJ…") as wh:
    try:
        event = wh.publish(
            "order.created",
            {"order_id": "ord_123", "total": 4200},
            idempotency_key="order-123",   # makes the publish safe to retry
        )
        print(event.event_uid, event.deliveries_created)
    except WebhooksAPIError as e:
        print(e.status_code, e.code, e.message)   # the stable {error:{code,message}} envelope

The API key comes from NimbusNexus Identity, not from webhookd — mint one in your NimbusNexus account console under API keys, choosing Webhooks as the product. Pick one scope: admin (full control), publish (produce events, read nothing) or read (read everything, produce nothing) — publish and read are disjoint, not a ladder — and optionally confine the key to a single project. The key is a JWT, so it starts with eyJ…; the SDK sends it as Authorization: Bearer <key>, exactly as before.

A whsk_… key no longer authenticates. webhookd no longer issues or accepts that key family, so an old whsk_… key is now refused like any unrecognised credential. Mint a replacement in the account console.

Transient failures (connection errors, 429, 5xx) are retried with backoff (a 429 honours Retry-After); other 4xx raise WebhooksAPIError.

Targeting a project

A project is addressed by its ID ("prj_3f9a…") — it has no slug or short name. Every project-aware call takes an optional project_id; leave it unset to target your workspace's default project, which is what a single-project workspace always wants:

wh.publish("order.created", {...})                            # -> the workspace's default project
wh.publish("order.created", {...}, project_id="prj_3f9a…")    # -> that specific project

Only the server can resolve "the default project" — the id is opaque and per-workspace, so there is no client-side name for it. Omitting the field is the way to ask for it; passing a made-up string ("default", a project's display name) is a 404. The id you need is on any response — event.project_id, endpoint["project_id"] — or from GET /v1/projects.

Outbox / durable buffering (producers)

publish() calls webhookd synchronously — if webhookd is unreachable it raises and the event is lost. The write-first outbox decouples the two: enqueue() durably persists the event to a pluggable Store and returns IMMEDIATELY (no network); drain() (or a background drainer) ships the buffered events later. Every send carries Idempotency-Key = record.id, so a re-drain after a crash or a lost response never double-publishes — webhookd dedupes. Delivery is at-least-once: nothing is lost while webhookd is down.

from nn_webhooks import Client, SQLiteStore

# 1. Configure a durable store (survives process restarts).
store = SQLiteStore("outbox.db")

with Client("https://webhooks.example.com", api_key="eyJ…", store=store) as wh:
    # 2. enqueue() instead of publish() — writes to the store and returns at once, NO network call.
    record_id = wh.enqueue("order.created", {"order_id": "ord_123", "total": 4200})

    # 3a. Drain on demand (returns {"sent", "failed", "remaining"}):
    wh.drain()

    # 3b. …or run a background drainer that calls drain() every 5s until the client closes.
    wh.start_drainer(interval_seconds=5)
    # ... your app keeps enqueuing; the drainer ships in the background ...
    wh.stop_drainer()   # also called automatically by Client.close()/__exit__

Idempotency guarantee. record_id is the idempotency_key you pass (or a generated UUID v4) and becomes the Idempotency-Key header on every delivery attempt for that record. If the process crashes after a send but before the response is recorded, the next drain() re-sends with the same key and webhookd returns the original event without re-fanning-out. A record that keeps failing is retried with capped exponential backoff up to max_attempts (default 10), then flagged dead (never retried again) and handed to the optional on_dead callback.

Built-in stores — pick one for Client(..., store=...):

The core SDK's only runtime dependency is httpx; RedisStore / PostgresStore lazily import their driver only when you construct them.

Schema change. The SQL stores' project column (a slug) is now project_id, nullable (null = the workspace's default project). There is no migration step — drain an outbox written by an older SDK before upgrading, or drop the webhookd_outbox table.

Manage endpoints & deliveries (operators)

The same Client wraps the control-plane API — register receivers and drain the dead-letter queue from code (needs an admin-scoped key). Management methods return the raw JSON as dicts (snake_case, exactly as the API sends); list methods return a page — {"items": [...], "next_offset": int | None}; delete_endpoint returns None (a 204).

from nn_webhooks import Client

wh = Client("https://webhooks.example.com", api_key="eyJ…")  # an admin-scoped key

# --- Endpoints ----------------------------------------------------------------
# Create a receiver — its signing secret is in the response exactly once, so persist it now.
ep = wh.create_endpoint(
    "https://your-app.example/webhooks",
    subscriptions=[{"match_kind": "prefix", "pattern": "order."}],
    description="orders service",
)
endpoint_id, signing_secret = ep["id"], ep["secret"]

wh.list_endpoints()                          # {"items": [...], "next_offset": ...} — default project
wh.list_endpoints(project_id="prj_3f9a…")    # …or scope the listing to one project by id
wh.get_endpoint(endpoint_id)

# PATCH — send only the keys you want to change (omitted = unchanged, None = cleared):
wh.update_endpoint(endpoint_id, {"max_attempts": 10, "status": "disabled"})

wh.rotate_endpoint_secret(endpoint_id)       # returns the new secret, once
wh.enable_endpoint(endpoint_id)              # recover an auto-disabled endpoint
wh.delete_endpoint(endpoint_id)              # -> None (204)

# --- API keys: these two methods no longer reach anything ---------------------
# They are still on the client and still import, but webhookd deleted POST and DELETE
# /v1/api-keys, so each call raises WebhooksAPIError with a 404 at runtime. Mint and revoke keys
# in your NimbusNexus account console instead.
wh.create_api_key("ci-publisher", scope="publish", expires_in_days=90)   # 404 — route is gone
wh.revoke_api_key("key_123")                                             # 404 — route is gone

# --- Deliveries / dead-letter recovery ----------------------------------------
for d in wh.list_deliveries(status="dead")["items"]:
    wh.redeliver(d["id"])

Develop

pip install -e '.[dev]'
pytest && ruff check . && mypy nn_webhooks