AshHooks

Webhooks for Ash Framework, in both directions:

• Inbound: verify webhooks from any sender, deduplicate them on a ledger (a table that records every delivery), and run your handler at least once with durable deduplication.
• Outbound: sign and deliver your own webhooks with retries, backoff, and dead-lettering through Oban or your own scheduler.

The two halves work independently: if you only receive webhooks, you need no queue infrastructure at all.

• Verify any sender. Describe a sender's signature scheme in a few lines of configuration: HMAC (hash-based message authentication code) with SHA-1, SHA-256, SHA-384 or SHA-512; RSA signatures, PKCS#1 v1.5 or RSA-PSS (the probabilistic signature scheme); ECDSA (elliptic curve digital signature algorithm) on P-256 or P-384; Ed25519; or a JWS (JSON Web Signature) body. The signature can sit in a header, a structured t=...,v1=... header, or the body itself, hex or base64 encoded, over any signed input, with timestamps, a replay window, and key rotation. When configuration cannot express a scheme, one callback can, written with the same public primitives.
• Duplicate and replayed deliveries are deduplicated on the ledger: exactly one row per delivery, and a crash mid-flight resumes on redelivery instead of losing the event (handlers run at least once).
• Outbound webhooks follow the Standard Webhooks spec, so receivers verify them with any conformant library. Key rotation and a legacy-envelope mode for receivers mid-migration are built in.
• Retries honor Retry-After, back off with jitter, dead-letter at a ceiling, and durably disable endpoints that return 410. A DNS failure is a transport error: a hostname that stops resolving retries, labeled nxdomain or resolve_timeout, and only a validated private or loopback answer dead-letters immediately.
• Safe defaults: inbound declarations use secret resolvers rather than literal binaries, and endpoint rows store secret references. Endpoint URLs are checked against server-side request forgery (SSRF) at registration and again at send time, and response bodies are never stored unless you opt in for a diagnostic run. Inbound ledgers retain decoded payloads and a digest of the signed bytes; outbound ledgers retain the exact bytes to send. Read access is yours to govern: the package injects no read policies (see Security).
• Telemetry events for the whole send and receive lifecycle, with classified reasons and identifiers rather than bodies or signing secrets.

Requires Elixir > 1.20 (OTP 28+) and Ash >= 3.34.3, < 4.0. Add Oban (> 2.20) for the generated delivery worker, and Phoenix or Plug for the HTTP ingress. Direct delivery and signature verification work without either dependency. The package follows semantic versioning over a named public surface; see Stability.

Installation

def deps do
  [
    {:ash_hooks, "~> 3.0"},
    # only for outbound delivery:
    {:oban, "~> 2.20"}
  ]
end

Or run mix igniter.install ash_hooks, which also tries to patch your endpoint's Plug.Parsers with a raw-body reader. Signature schemes sign the exact wire bytes, and a router plug cannot recover what the parser already consumed, so if the automatic patch didn't apply, add it yourself:

plug Plug.Parsers,
  parsers: [:json],
  pass: ["*/*"],
  body_reader: {AshHooks.BodyReader, :read_body, []},
  json_decoder: Phoenix.json_library()

By default every parsed request carries a cached copy of its raw body; pass [only: ["/webhooks"]] as the reader's third element to limit that memory cost to your webhook routes.

You also create your own tables: ash_hooks injects fields and identities onto your resources, and your migrations carry them, including the unique indexes the deduplication guarantee rests on. Complete, runnable migrations, resources, and setup are in the get-started tutorial.

Upgrading from 2.x? 3.0 removes the built-in vendor verifiers and requires an explicit provider on every inbound declaration. UPGRADING.md gives the replacement configuration for each removed module.

Receiving webhooks

1. Describe the sender

A provider module describes how one sender signs its requests. For most senders, configuration is all it takes:

defmodule MyApp.Webhooks.Payments do
  use AshHooks.Provider,
    verify: [
      algorithm: :hmac_sha256,
      signature: {:structured_header, "x-payments-signature", signature_key: "v1"},
      timestamp: {:signature_field, "t"},
      signed_input: [:timestamp, ".", :body],
      tolerance_seconds: 300
    ]
end

This one reads x-payments-signature: t=1731312000,v1=5257a8..., checks the lowercase-hex HMAC-SHA256 of timestamp.body in constant time, accepts any listed v1 signature (so the sender can rotate keys), and rejects a valid signature whose timestamp is more than 300 seconds from the clock. Options are checked when the module compiles.

The Verify any sender guide has recipes for common shapes: an HMAC of the body in a header, a prefixed signature, hex or base64 secrets, Standard Webhooks, signed method and URL, sorted form fields, RSA and ECDSA headers, Ed25519, JWS bodies, custom timestamp formats, and two schemes on one route. AshHooks.Provider.Scheme documents every option. When a scheme needs something configuration cannot express, implement verify_signature/3 with the public primitives in AshHooks.Provider.Signature.

ash_hooks ships no vendor-specific verifiers. A sender's documentation is the authority on its scheme, and a scheme change on the sender's side never waits on an ash_hooks release.

2. Declare the source on a ledger

use Ash.Resource,
  data_layer: AshPostgres.DataLayer,
  extensions: [AshHooks, AshHooks.InboundDelivery]

inbound_delivery do
  # sender event ids aren't unique across accounts: your scope slots extend
  # the dedup identity (each must be a non-nullable attribute)
  scope_identity([:account_id])
end

attributes do
  attribute(:account_id, :string, allow_nil?: false)
end

webhooks do
  inbound :payments do
    provider MyApp.Webhooks.Payments
    secret {:app_env, [:my_app, :payments_webhook_secret]}
  end
end

Secrets are always sources, never literal values: an {m, f, a} callback, an {:app_env, path}, a zero-arity function, or (for multi-tenant apps) a one-arity function that receives the tenant. For public-key schemes the source returns the sender's public key, PEM-encoded.

The inbound name (:payments) identifies the source in the ledger. provider is required.

3. Ingest from your controller

One call runs the whole pipeline: verify the signature over the raw bytes, persist the payload, deduplicate, claim under a lease (a time-limited ownership claim; rows whose worker died are reclaimed when it expires), run your handler, and record the outcome.

case AshHooks.Ingress.ingest(Ledger, :payments, conn.private[:ash_hooks_raw_body], %{
       headers: conn.req_headers,
       scope: %{"account_id" => conn.params["account_id"]}
     }) do
  # Acknowledge terminal rows, including permanent failures.
  # A duplicate may have retried its handler: inspect its current status.
  {:ok, _tag, %{status: status}} when status in [:processed, :failed_permanent, :superseded] ->
    send_resp(conn, 200, "")
  {:ok, _tag, _row} -> send_resp(conn, 500, "")    # handler failed: the sender retries
  # not authentic, a replay, or an identity the provider rejected: do not retry
  {:error, %AshHooks.Errors.Invalid.InvalidSignature{}} -> send_resp(conn, 400, "")
  {:error, %AshHooks.Errors.Invalid.StaleTimestamp{}} -> send_resp(conn, 400, "")
  {:error, %AshHooks.Errors.Invalid.MalformedPayload{}} -> send_resp(conn, 400, "")
  # paused source, missing secret, storage failure: the sender retries
  {:error, _} -> send_resp(conn, 503, "")
end

When a scheme signs the HTTP method or URL, also pass method: conn.method and request_uri: the public URL. Build it from a base URL you configure, not from conn, which behind a TLS proxy carries the internal host and port.

4. Handle the event

A configured provider classifies every verified request as :webhook and returns an AshHooks.Provider.Event holding the payload. Override parse_event_type/1 and handle_event/2 in the same module to map the sender's event names to your own atoms and run your application's actions:

defmodule MyApp.Webhooks.Payments do
  use AshHooks.Provider, verify: [...] # the options shown in step 1

  @types %{"charge.succeeded" => :charge_succeeded}

  @impl true
  def parse_event_type(%{"type" => type}) do
    case Map.fetch(@types, type) do
      {:ok, atom} -> {:ok, atom}
      :error -> {:error, :unknown_event_type}
    end
  end

  def parse_event_type(_payload), do: {:error, :malformed_payload}

  # return {:ok, result} or {:error, :retry | :permanent, reason}
  @impl true
  def handle_event(:charge_succeeded, payload), do: MyApp.Billing.record_charge(payload)
end

Map event names with an explicit table; never call String.to_atom/1 on request data. An unknown type is recorded as a permanent failure.

Delivery semantics. A terminal delivery (:processed, :failed_permanent, or :superseded) is never processed again. A crash after your handler ran but before the ledger recorded it will re-run the handler on redelivery, so write handlers idempotent, keyed on the logical event or batch identity (for action-level idempotency elsewhere in your app, the sibling package ash_onetime is an optional companion). Permanent failures belong on your operator surface; the successful acknowledgment prevents a sender from retrying a row that cannot run again.

The deduplication identity comes from the declaration's event_id extractor, then the provider's optional event_identity/1 callback, then the SHA-256 of the raw body.

claim_delivery/3, mark_processed/4, and friends are public if you want to drive the lease machine from your own async pipeline; AshHooks.Ingress.reap/2 re-drives deliveries whose claims died with an expired lease.

Payload formats

JSON is the default. Opt into flat application/x-www-form-urlencoded decoding on a route with body_format :form, or implement the provider's optional body_format/0 callback returning :form. Names and values are URL-decoded; bracket notation remains literal, and repeated names keep their last value in the stored map. Invalid UTF-8 names or values become a permanent malformed payload, matching JSON's failure handling. Verification always receives the original raw bytes, before decoding. The request's Content-Type cannot opt a route into a decoder.

inbound :sms do
  provider MyApp.Webhooks.Sms
  body_format :form
  secret {:app_env, [:my_app, :sms_webhook_secret]}
end

Providers may implement decode_payload/1 for a signed envelope (a configured JWS-body provider generates one); an explicit route body_format overrides it.

Rotation, pausing, and inbound queries

• previous_secret on an inbound declaration takes the outgoing key's source: both keys verify until you remove the option. Tenant-aware (one-argument) previous_secret and paused callbacks receive the resolved tenant on multitenant ledgers.
• paused refuses requests with {:error, :source_paused} before verification or any write; answer 5xx and the sender retries, so nothing is lost.
• AshHooks.Ingress.list/2 queries inbound deliveries by source, status, event type, and time range; counts/2 answers a window's received, duplicate, failed-verification, and handler-failure counts (ingress_tracking :on).

HTTP Message Signatures with RequestSeal

For senders that sign with HTTP Message Signatures (RFC 9421), use RequestSeal 1.0 or later ({:request_seal, "~> 1.0"}) alongside ash_hooks. It verifies which parts of the request are covered (method, authority, path, query, and named headers), checks signature freshness and expiry, verifies body integrity with content digests (RFC 9530), and spends each signature's nonce once through a replay store you choose (ETS, Erlang Term Storage; PostgreSQL; or ash_onetime). Keys come from a resolver you write, and its bounded key discovery handles JSON Web Key Sets (JWKS) and key directories; it also verifies Web Bot Auth agents and crawlers. Its RequestSeal.AshHooks adapter verifies inside an ash_hooks provider, and RequestSeal.AshHooks.Http signs your outbound ash_hooks deliveries with RFC 9421. ash_hooks keeps the ledger, retries, and transport. The Verify any sender guide shows the provider.

Sending webhooks

Declare the event on the emitting resource, point it at your subscription and delivery resources, and define one worker module:

defmodule MyApp.WebhookDeliveryWorker do
  use AshHooks.Worker,
    deliveries: MyApp.OutboundDelivery,
    endpoints: MyApp.WebhookEndpoint,
    secret_resolver: {MyApp.Secrets, :webhook_secret},
    queue: :webhooks
end

The secret resolver maps an endpoint's secret reference to its value: generate values with AshHooks.Signing.generate_secret/0, store them whole in your secret store, and return them unchanged.

Related budgets are checked at compile time: the macro refuses an Oban :max_attempts at or below :delivery_max_attempts (only the row's ceiling decides dead-letter), a literal http_opts timeout: at or above :attempt_timeout, non-positive budget values, and a literal validate_destination: false outside :test (it is the loopback test seam). Timing is also configurable at runtime: pass runtime_config: {MyApp.Config, :webhook_runtime} and the callback resolves :delivery_max_attempts, the backoff ladder, the Retry-After cap, :attempt_timeout, and :finalization_allowance per perform; every compile-time misfit check re-runs against the resolved set, so an override that overruns the baked job envelope fails the job before any delivery work. See AshHooks.Worker for every relation and its rationale.

Dispatch, wiring the worker's generated enqueue function:

{:ok, event} =
  AshHooks.Event.new(
    type: :order_paid,
    payload: Jason.encode!(%{order_id: order.id}),
    # Reuse this ID when retrying the same logical event.
    id: "msg_order-" <> to_string(order.id)
  )

AshHooks.dispatch(Order, :order_paid, event,
  enqueue: {MyApp.WebhookDeliveryWorker, :enqueue}
)

Every matching enabled endpoint gets a durable delivery row carrying the exact bytes to sign. The worker signs per Standard Webhooks (the same webhook-id on every retry), succeeds only on 2xx, never follows redirects, honors Retry-After (bounded) on 408/429/5xx, backs off with jitter when the header is absent (5xx, transport errors), dead-letters other client errors, and durably disables the endpoint on 410. An endpoint's failure never blocks delivery to its siblings.

Without Oban, dispatch still works (every matching endpoint gets a durable :pending row), but nothing drives those rows until you define the worker (or call AshHooks.Delivery.run/2 yourself): the delivery row owns the retry policy, and the queue is only its trigger (architecture decision record ADR-0008).

The default HTTP adapter bounds response headers, retained body bytes, and the complete operation time. OTP's :httpc is an alternative with additional buffering limits described in its module docs. You can also provide an adapter for your application's transport requirements.

Optional outbound policy

All three options are off by default. The worker and direct delivery runner accept:

• retryable_statuses: [408, 409, 425, 429, 500, 502, 503, 504] replaces the retryable HTTP-status set. Omit it for the default (408, 429 and 5xx); use [] to make every ordinary HTTP failure final. Retry-After, backoff, and ceilings apply to configured retry statuses. Success on 2xx, redirect refusal, and the durable 410 disable rule remain fixed. Transport errors keep the default retry policy.
• delivery_headers: {MyApp.Webhooks, :headers} calls headers(delivery, endpoint, tenant) before each send. Return a header map or list of binary pairs. Direct-run configuration also accepts that map/list or a three-argument function. Names are checked case-insensitively: hosts cannot override webhook-id, webhook-timestamp, webhook-signature, x-webhook-signature, Content-Type, Host, or HTTP framing/connection headers. Host metadata such as x-webhook-event, trace IDs, and dates is allowed. Invalid headers and callback failures retry without sending; their values never become error summaries.
• certificate_pin: {MyApp.Webhooks, :pin} calls pin(endpoint, tenant) so a host can choose a pin per endpoint. Return nil for off, {:certificate, der} for the exact leaf certificate, {:sha256, digest} for its SHA-256, or {:spki_sha256, digest} for the DER SubjectPublicKeyInfo SHA-256. Digests are 32 raw bytes. Direct configuration also accepts the pin or a two-argument function. A static host pin can be supplied in http_opts[:certificate_pin]; a configured resolver takes precedence, including an explicit nil result. The pin requires HTTPS and AshHooks.Http.Bounded; it adds to normal CA, hostname/IP-SAN, SSRF, and deadline checks. Mismatch, invalid pin, or resolver failure closes the attempt without a request and records a non-retryable failure. Default: off. Load computed worker pins through the callback or the existing http_opts: {module, function, args} runtime seam.

All callbacks resolve from current host configuration on each attempt. They need no tables, endpoint attributes, process-local state or extra dependencies.

Schedule AshHooks.reconcile_pending/3 for each declaration and tenant using the same named worker callback. It repairs enqueue gaps, due retries, expired send leases, and pending endpoint disables while preserving retry times and attempt counts. Delivery rows retain their declaration and route, so one declaration cannot recover another's work. Anonymous enqueue callbacks can use a stable enqueue_key for recovery. Application nodes must keep UTC clocks synchronized for lease decisions.

Each result belongs to a live, fenced attempt. Network delivery remains at-least-once: a receiver may accept a request before its sender dies. Receivers deduplicate by the stable webhook-id.

Delivery logs and attempt history. Add the AshHooks.OutboundAttempt extension on a resource and pass attempts: to the worker: every attempt records one durable row (time, outcome, classified reason, HTTP status, latency, the signing key reference but never the secret), queryable by endpoint, outcome, event type, and time range with keyset pagination (AshHooks.Delivery.list/2 and AshHooks.OutboundAttempt.list/2). Response bodies stay off unless an endpoint opts in per row (response_capture_attributes :on + response_capture: true): captured bytes are bounded, redacted, and only allowlisted response headers are kept.

Response bodies are never stored by default: each delivery row keeps the status and a content-type summary. For a diagnostic run, enable capture on one row. The captured body is bounded, passes built-in redaction, and is marked [captured] in the snippet:

# Diagnostic capture for one row. Supply the resource and resolver config;
# retry overrides are independent of the worker's configured settings.
AshHooks.Delivery.run(
  %{"endpoint_id" => row.endpoint_id, "event_uuid" => row.event_uuid},
  snippet_capture: true,
  deliveries: MyApp.OutboundDelivery,
  endpoints: MyApp.WebhookEndpoint,
  secret_resolver: {MyApp.Secrets, :webhook_secret},
  max_attempts: 10, base_backoff_seconds: 2,
  max_backoff_seconds: 3600, retry_after_cap_seconds: 86_400
)

The driver defaults to the retry values shown. Its max_attempts option controls the delivery row; the worker calls that option delivery_max_attempts because the worker's max_attempts controls Oban jobs. Copy the worker's retry policy for a diagnostic run: a lower delivery ceiling can dead-letter a row the worker would still retry.

Only non-terminal rows are driven: a row that already finished will not re-send; re-drive a failed one, or wait for its retry.

For a domain-specific denylist, also pass a snippet_redactor ({module, function} or fn body -> body | nil; return the redacted binary, or nil to leave the body uncaptured). It runs before the built-in redaction; a crashing redactor leaves the body uncaptured. See AshHooks.Worker for the worker-macro form.

Test events. AshHooks.Delivery.test_send/2 sends a signed synthetic event of a chosen type to one endpoint, outside subscriptions: a fresh webhook id each time, a payload that marks itself as a test, and the normal attempt/telemetry lifecycle behind it (see AshHooks.Delivery).

Signing modes. :standard (default) needs only the endpoint's secret_ref. :dual and :legacy additionally require a legacy_secret_ref: :dual emits both envelopes so receivers can migrate, :legacy emits only the old one.

Secret rolling. Roll an endpoint's signing secret with an overlap window (AshHooks.Endpoint.roll_secret/2 + key_lifecycle :on): the incumbent reference moves to previous_secret_ref, BOTH keys sign every delivery until the window ends (Standard Webhooks multi-signature), the scheduled sweep retires the old key on time, and the row exposes each key's created, first-used, and retired times, never secret material.

Endpoint health and auto-disable. AshHooks.EndpointHealth.stats/2 computes one endpoint's trailing-window success rate, latency percentiles, and first-attempt vs eventual successes from the attempt history; the endpoint's auto_disable policy (a duration and failure share beside the existing 410 rule) disables sustained failures through the same durable switch, warning first ([:ash_hooks, :endpoint, :disable_warning]) so your notify surface can reach people before the breaker opens. Re-enabling is explicit.

The event catalog and version pinning. The outbound DSL declares each event type's description, payload_schema, example, and version; AshHooks.Catalog reads the catalog back for event pickers and doc generation. An endpoint that pins a payload version (version_pinning :on) receives that variant builder's exact bytes; per-endpoint rows carry their own version.

The event log. AshHooks.OutboundEvent records each dispatched event once (id, type, version, exact bytes) whether or not any endpoint was enabled, with its own retention, and replay/3 delivers one event or a filtered set to any chosen endpoint, including one added later, keeping the event id so receivers deduplicate.

Subscription attribute filters. filter_attributes([:order_id, :region]) on the subscription resource lets a subscription match on declared event attributes (values ride the event's metadata), evaluated before any delivery row exists: multiple filters AND, list values any-of, unknown keys never match.

Transports. AshHooks.Transport routes one endpoint's deliveries to your own transport (a developer's local forwarder behind a CLI listen command) instead of HTTP, with the same signing, attempts, and telemetry. A destination policy (AshHooks.DestinationPolicy) adds your own allow and deny rules around the packaged SSRF check.

Fitting the extensions to your domain

The injected fields and actions carry opinionated names by default; your domain may reserve those names or own the lifecycle itself. Every knob is a per-resource DSL option, fail-closed at compile:

# your domain reserves `payload` for its own sole payload store:
# rename the ledger's exact-bytes column instead
outbound_delivery do
  payload_attribute :event_bytes
end

# append-only audit ledger: no destroy action is injected, and
# the retention hook fails loud (deletion is your own surface)
outbound_delivery do
  payload_attribute :event_bytes
  prune_action :none
end

# your register already has an enable switch: map the durable
# enable/disable onto it (the 410 breaker and operators flip your
# attribute; the package injects no `status` of its own)
endpoint do
  status_attribute :active
  enabled_values [true]
  disabled_value false
end

Ledger operations use the complete resource primary key, including custom and composite keys. Every component needs a default, a data-layer generator, or a supported create input; inbound scope fields can supply key components. Endpoint and subscription references require a single UUID-compatible key, which may be renamed or generated with uuid_v7_primary_key. A subscription register may be a closed {:array, :atom} enum (your constraints, your default; matching handles atoms and strings identically, wildcard included), and injected-PK ledgers still classify :created/:duplicate exactly.

Policies. The package injects write actions, not the authorization around them. Cover the generated machine actions with an action-specific policy of your own (the runtime's internal calls bypass policies by design), or keep the resources off every actor-facing surface; the full action set and obligation are stated at each extension site in the module docs.

Observability

Attach one handler to see the whole lifecycle: inbound verify/dedup/claim, enqueue failures, delivery attempt/result/backoff/dead-letter/endpoint-disable. Events carry ids, integers, fixed atoms, and classified reasons, never secrets, bodies, or payloads. The exact event list and a copy-paste attach_many block are in the AshHooks.Telemetry docs and the get-started tutorial.

Retention

Ledger and delivery rows accumulate by default (they are the dedup and audit record). When you want them bounded, drive the retention hooks on a schedule of your choosing (an Oban cron job, a mix task, a nightly job):

• AshHooks.Ingress.prune/2 and AshHooks.Delivery.prune/2 delete terminal rows older than a cutoff; retryable and in-flight rows are never touched. They key off the resource's inserted_at, so add Ash's timestamps() to the resource and its migration. Append-only ledgers may omit the destroy action entirely (prune_action :none); prune/2 then fails loud with a named error, and deletion is your own surface.
• AshHooks.Ingress.redact_payload/5 rewrites a claimed row's payload under the claim fence (scrub sensitive fields while keeping the dedup identity; the original-bytes digest is preserved for audit).

Deleting a terminal row re-opens its dedup identity (a replayed webhook re-processes, a re-emitted outbound event re-sends), so set the TTL beyond any replay or re-emission horizon.

Multi-tenancy

If one app serves multiple organizations, ash_hooks works the way Ash does: declare attribute multitenancy on the four resources (your Subscription, Endpoint, inbound ledger, and outbound delivery ledger) and pass a :tenant to every call:

# the four resources declare the same contract
multitenancy do
  strategy :attribute
  attribute :org_id
end

# outbound: the tenant scopes the fanout, endpoint resolution, and rows
AshHooks.dispatch(Order, :order_paid, event, tenant: org.id)

# inbound: the tenant rides the request context
AshHooks.Ingress.ingest(WebhookLedger, :payments, raw_body, %{
  headers: conn.req_headers,
  tenant: org.id
})

From there, isolation is enforced by the data model: a dispatch for tenant A cannot create a delivery row for tenant B's endpoint, a subscription pointing at another tenant's endpoint id resolves to not-found and is skipped, the inbound dedup identity is per-tenant, and the claim fence and every mark are per-tenant. Touch a multitenant resource without a tenant and you get {:error, :tenant_required} before any data access; mix tenancy declarations across the resources one operation touches (say, a tenant-scoped delivery ledger beside an undeclared endpoint table) and you get {:error, :tenancy_mismatch}. Both are named errors, and both fail closed. Each operation checks the set it actually reads: an inbound-only app declares tenancy on its ledger alone and needs none of the outbound resources.

The async path is tenant-aware too: the Oban job args include the row's tenant (the worker recovers full context after any restart), the retention sweeps take a :tenant (with AshHooks.reap_all/2 / AshHooks.prune_all/2 sugar to sweep a tenant list), and AshHooks.reconcile_pending/3 repairs rows stranded between the row write and the enqueue, per tenant, with single-winner semantics. If your signing secrets are per-tenant, resolution can be too: the worker macro's tenant_aware_secrets: true (the resolver becomes f(ref, tenant)), an inbound secret fn tenant -> {:ok, secret} end, and an optional provider webhook_signing_secret(connection, tenant) callback.

Single-tenant app? Do none of this. No tenant passed means no tenant threaded; behavior is identical.

Adopting tenancy on tables that already have rows is an ordered transition (backfill, then regenerate indexes, then enable); the adoption checklist walks it.

Security

Signatures are compared in constant time. Secret values come from your configured sources. Endpoint URLs are checked at registration and send time; the supplied adapters connect to a validated address to prevent DNS rebinding. The default adapter bounds HTTP reads, and diagnostic response capture runs through redaction before storage.

Your resource policies govern access to the ledgers. Inbound rows contain decoded provider payloads, signed-body digests, event IDs, and scope keys; outbound rows contain exact event bytes. Those payloads can contain personal information. Define read policies that deny access by default:

policies do
  # deny by default; open exactly what your app needs
  policy action_type(:read) do
    authorize_if actor_attribute_equals(:admin, true)
  end
end

This assumes Ash.Policy.Authorizer in the resource's authorizers; match the snippet to your actual actors. The policies block lives inside the ledger/delivery resource module (with authorizers: [Ash.Policy.Authorizer] in the use Ash.Resource options; Ash's policies guide covers the full model). Vulnerability reports: SECURITY.md, never a public issue.

Stability

ash_hooks follows semantic versioning over a named public surface (the DSL, the public modules, injected attributes/actions, telemetry events, error classes). The current major is 3: breaking changes next ship in 4.0, deprecations run at least two minor releases, and safety corrections ship as fixes (ADR-0010).

Minimum supported versions: Elixir ~> 1.20 (OTP 28+; CI-tested on Erlang/OTP 28 and 29), Ash >= 3.34.3 and < 4.0, Oban ~> 2.20 (optional, outbound only). Development is supported on macOS and Linux; Windows developers use WSL2. CI runs on Linux, including an AshPostgres leg exercising a uuid_v7-keyed consumer shape. On Ash 3.33+ your application must also set Ash's required default_string_length_count config: an Ash requirement for every app compiling resources, not an ash_hooks one (UPGRADING.md). One nuance: a fix that closes a safety hole can change behavior in a patch release (a delivery that wrongly succeeded may now retry, for example); such corrections are always called out under "Fixed" in the CHANGELOG. Moving off a supported version is a minor release with an UPGRADING.md note.

Further reading

• Get started: a complete walkthrough, migrations included
• Guided tour (Livebook): run the whole library inside one notebook
• Verify any sender: signature recipes and the provider configuration reference
• UPGRADING.md: migration notes per release
• SECURITY.md: reporting and scope
• DSL reference
• Architecture decisions: the reasoning behind every major design choice
• CONTRIBUTING.md: setup, the gate suite, and how to propose changes; GitHub issues are the support channel
• Full API docs at hexdocs.pm/ash_hooks

License

MIT.