@giveitsmaller/contracts 0.68.0 → 0.70.0

package/openapi/api.yaml

@@ -89,7 +89,7 @@ info:
     of truth instead of hardcoding magic numbers. A runtime
     `GET /api/uploads/limits` endpoint for dynamic discovery
     (per-tier / per-environment overrides) is a deferred follow-up.
-  version: 2.198.0
+  version: 2.201.0
   contact:
     name: API Support
 
@@ -454,10 +454,26 @@ paths:
                     feature_tier_restricted: '#/components/schemas/FeatureTierRestrictedResponse'
         '413':
           description: |
-            File exceeds the absolute maximum upload size (across all
-            tiers). For tier-specific size caps below this absolute
-            limit, the server returns 403 with
-            `error_type: tier_restriction, restriction_kind: file_size`.
+            Two rejection cases, discriminated by the `ErrorEnvelope.error`
+            code — the same pair as `POST /api/uploads`.
+
+            - **Byte size** — the file exceeds the absolute maximum upload size
+              (across all tiers). For tier-specific size caps below this
+              absolute limit, the server returns 403 with
+              `error_type: tier_restriction, restriction_kind: file_size`.
+            - **Pixel dimensions** (`error: IMAGE_DIMENSIONS_TOO_LARGE`) — as
+              declared on `POST /api/uploads`: a decodable raster image whose
+              `width x height` exceeds the configured pixel ceiling
+              (`UPLOAD_MAX_IMAGE_PIXELS`, default 16 MP), read from the file
+              **header** and rejected **before** decoding.
+
+            🔴 **THIS CASE WAS UNDECLARED HERE UNTIL 2026-08-28, AND THIS IS
+            THE PATH IT MATTERS MOST ON.** Multipart exists for LARGE uploads,
+            so a high-pixel image is *more* likely to arrive here than on the
+            single-shot endpoint where the ceiling was documented. A ceiling
+            declared on the small-file path and omitted from the large-file
+            path is worse than one declared nowhere: a consumer who reads the
+            contract concludes the rule does not apply here.
           content:
             application/json:
               schema:
@@ -1244,20 +1260,29 @@ paths:
         - name: archived
           in: query
           required: false
-          x-availability: planned
-          description: |
-            🔴 **NOT IMPLEMENTED — see the `planned` note below. A server
-            that does not support this parameter SILENTLY IGNORES IT and
-            returns the DEFAULT (non-archived) list.** That is the dangerous
-            half of this feature and the reason it is called out here rather
-            than only on the endpoints: `POST .../archive` would 404, which is
-            loud and teaches the caller immediately, whereas
-            **`archived=true` returns a 200 carrying the wrong rows.** A
-            consumer asking for their archive receives their NON-archived
-            workflows and is told nothing. **Silent wrong data beats a 404
-            every time for damage.**
-            Until the server implements it, a client MUST NOT present an
-            "archived" view built on this parameter.
+          x-availability: stable
+          description: |
+            ⚠️ **A SERVER THAT DOES NOT SUPPORT THIS PARAMETER SILENTLY
+            IGNORES IT and returns the DEFAULT (non-archived) list** — it
+            does not error. That asymmetry is why this is called out on the
+            parameter and not only on the endpoints: `POST .../archive`
+            would 404, which is loud and teaches the caller immediately,
+            whereas **`archived=true` returns a 200 carrying the wrong
+            rows** — a consumer asking for their archive receives their
+            NON-archived workflows and is told nothing. **Silent wrong data
+            beats a 404 every time for damage.**
+
+            ✅ **IMPLEMENTED. This was `planned` and carried a "clients MUST
+            NOT present an archived view" prohibition; both are struck.**
+            Verified on `compression_api` `origin/main` before the flip
+            rather than on report: the `ArchiveWorkflow` and
+            `RestoreWorkflow` command handlers exist, and
+            `WorkflowListReadPort` takes the `bool $archived` filter.
+            ⚠️ **What that establishes is that the code is MERGED, not that
+            a given environment is running it** — availability in this
+            contract has no environment axis, so read this as the shape
+            being committed, and confirm deployment from the environment
+            rather than from here.
 
             Archived-row filter (ticket
             [`j2s6T2qf`](https://trello.com/c/j2s6T2qf)). **Default `false`**:
@@ -5827,7 +5852,29 @@ paths:
       summary: Get current credit balance
       description: |
         Returns a snapshot of the caller's credit position at request
-        time. Per ticket [I23 `DffjC3zm`](https://trello.com/c/DffjC3zm)
+        time.
+
+        🔴 **EVENTUALLY CONSISTENT WITH A COMPLETED CHECKOUT, AND THIS
+        RESPONSE CARRIES NO FRESHNESS SIGNAL.** A credit-pack purchase is
+        granted on an asynchronous server-to-server callback, so a
+        snapshot taken just after the Stripe redirect **may legitimately
+        predate the grant**. There is **no timestamp, no version, no
+        pending state and no session correlation** here — the fields
+        below are the whole of it — so **this endpoint cannot
+        distinguish "the grant has not landed yet" from "the payment did
+        not credit you".**
+
+        ⇒ **Do not treat an unchanged balance after checkout as a
+        failure.** Those two states need opposite handling — *wait*
+        versus *contact support* — and the way to tell them apart is
+        `GET /api/v2/credits/usage`, correlating on the checkout
+        `session_id`, not this endpoint. See
+        `POST /api/billing/checkout` for the full model, **including
+        that nothing detects a callback which never arrives.**
+
+        ⛔ **No interval is stated anywhere in this contract**, because
+        none has been measured end to end. Any client timeout is the
+        client's own choice and is not sanctioned here. Per ticket [I23 `DffjC3zm`](https://trello.com/c/DffjC3zm)
         + plan v5 §F9 round-13 narrowing — this endpoint is the
         canonical user-visible billing-state surface.
 
@@ -6232,6 +6279,58 @@ paths:
         frontend can hide/disable the upgrade affordance without
         hard-coding the rollout state.
 
+        🔴 **THE CREDIT GRANT IS ASYNCHRONOUS. THE BALANCE WILL NOT
+        REFLECT IT WHEN STRIPE REDIRECTS THE USER BACK.** Credits are
+        granted on the server-to-server callback, which arrives
+        independently of the browser redirect. A client reading
+        `GET /api/v2/credits/balance` immediately on return **may
+        legitimately see the pre-purchase value, and that is not an
+        error.**
+
+        **How to observe the grant:** poll `GET /api/v2/credits/usage`
+        for a transaction whose `reference_id` equals the `session_id`
+        returned here, and whose `reference_type` is
+        `stripe_checkout_session`. That is the only client-observable
+        correlation between a checkout and its credits.
+
+        🔴 **MATCH ON `reference_type` + `reference_id`, AND DO NOT
+        FILTER ON `type`.** The pack grant is written as an
+        **`adjustment`** row, not a `top_up` — a client filtering for
+        `type: top_up` finds nothing however long it polls. `type` is a
+        free-form string here, and the pack path's value is the one
+        named in this sentence.
+
+        ⚠️ **This block previously said the pack path did not populate
+        `reference_id` "yet".** It does. That sentence told a client the
+        correlation was unavailable, so a client believing it built no
+        correlation at all and fell back to guessing from timing or from
+        description text — **the exact ambiguity the field exists to
+        remove.** A stale "not yet" retires a capability that shipped,
+        and this direction of staleness is invisible: nothing fails, the
+        client simply never asks.
+
+        ⛔ **NO TIMING IS STATED, DELIBERATELY.** The interval is
+        dominated by the provider's callback delivery latency, which
+        **this system does not measure**. Production has processed
+        **zero** such events; staging exactly one, a synthetic rehearsal
+        purchase — n=1, idle box, caller not a browser. **No bound is
+        derivable from that**, and a client MUST NOT build a timeout
+        from any figure in this contract, because there is none.
+
+        ⇒ **Show a pending state rather than a failure.** Absence of
+        credits after checkout is **indeterminate, not negative**.
+
+        🔴 **AND THE BRANCH A CLIENT MUST NOT BE PROMISED AWAY FROM:
+        NOTHING DETECTS A CALLBACK THAT NEVER ARRIVES.** Recovery relies
+        on the provider's own retries. There is **no server-side
+        reconciliation** comparing completed checkouts against granted
+        credits, and an event already marked terminal **cannot be
+        reopened**. **No customer-visible recovery route exists and no
+        automated one** — the manual path is an operator action. ⇒ A
+        client that has waited past any interval it considers reasonable
+        should **direct the user to support with the `session_id`**,
+        which is the only handle both sides share.
+
         The Stripe **webhook** (`POST /api/webhooks/stripe`) is a
         server-to-server, signature-authenticated callback and is
         deliberately **not** part of this client contract.
@@ -6945,6 +7044,10 @@ components:
         customer path carries the same signal as the SSE/message
         surface. Cross-spec enum parity is verified by
         `tests/test_asyncapi_named_schemas.py`.
+      # OPEN. Purely diagnostic — it reports which path `auto` took so a caller can
+      # see why. An unrecognised path does not change any decision a consumer makes,
+      # so a new one is safely ignorable and additive.
+      x-enum-openness: open
       enum:
         - stream_copy
         - re_encode
@@ -6971,6 +7074,12 @@ components:
         See `schemas/FORMAT.md` §Availability Taxonomy and
         `docs/decisions/0001-contract-first-availability.md` for the full
         rules, vocabulary rationale, and parser obligations.
+      # CLOSED. A consumer uses this to decide whether to EXPOSE a feature, so an
+      # unrecognised level cannot be handled safely in either direction — hiding a
+      # live capability or surfacing one that is not. ADR-0001 §1.3 defines exactly
+      # these five; a sixth is a contract event that consumers must be updated for,
+      # and failing loudly is the correct way to require that.
+      x-enum-openness: closed
       enum:
         - stable
         - beta
@@ -7873,8 +7982,23 @@ components:
         overridden:
           type: boolean
           description: |
-            True iff a per-account override is active (i.e. `effective`
-            comes from an override rather than the tier default).
+            Whether `effective` came from a per-account override rather than
+            the tier default.
+
+            🔴 **RESERVED, AND STRUCTURALLY `false` TODAY — DO NOT BRANCH ON
+            IT.** No per-account override store exists, and the API's resolver
+            is called with no override for either limit, so **every response
+            carries `false`, and will until an override store ships.**
+            Measured by `compression_api` on 2026-08-28.
+
+            ⚠️ It is declared **required**, so a consumer must handle a field
+            that currently takes exactly one value. It is kept rather than
+            removed because this is the shape an environment-overridable limit
+            needs, and removing it would be a breaking change to re-make later.
+
+            📌 **A test asserting `overridden === false` would be a decoration
+            — no input makes it fail.** When an override store exists, assert
+            against that, not against this field now.
 
     CreditTransaction:
       type: object
@@ -8001,10 +8125,44 @@ components:
           type: string
           description: |
             Identifier of the referenced entity. Typically a UUID v7
-            for `workflow` references, but other reference types
-            (`top_up` carrying a Stripe payment intent ID,
-            `cycle_grant` carrying a cycle date, etc.) may use
-            non-UUID identifiers. NOT constrained to `UuidV7`.
+            for `workflow` references; other reference types may use
+            non-UUID identifiers and this is NOT constrained to
+            `UuidV7`.
+
+            🔴 **FOR A CREDIT-PACK PURCHASE THIS IS THE CHECKOUT
+            SESSION ID (`cs_...`) — THE IDENTIFIER THE CLIENT ACTUALLY
+            HOLDS**, paired with a `reference_type` of
+            `stripe_checkout_session`. It is returned as
+            `BillingCheckoutSession.session_id` when the session is
+            created, and it is the handle that makes the asynchronous
+            credit grant observable. See `POST /api/billing/checkout`.
+
+            🔴 **THE ROW IS AN `adjustment`, NOT A `top_up`.** Filter on
+            `reference_type` — not on `type` — or the poll described on
+            checkout matches nothing.
+
+            ⚠️ **Populated only when the provider callback carries a
+            session id.** It is written straight from that callback, so
+            a callback without one leaves both `reference_type` and
+            `reference_id` `null` rather than inventing a correlator.
+            Treat absence as "not correlatable", never as "not granted".
+
+            ⚠️ **This description previously named a Stripe PAYMENT
+            INTENT id.** A client never receives one — checkout returns
+            `checkout_url` and `session_id` and nothing else — so the
+            only correlator this contract named was one the caller
+            could not obtain.
+
+            ⚠️ **AND IT THEN SAID THE PACK PATH LEFT THIS FIELD EMPTY
+            "until the session id is wired through".** It is wired:
+            measured on `compression_api` `origin/main` (`8c136261`),
+            `CheckoutSessionCompletedHandler::fulfillPack` passes
+            `referenceType: 'stripe_checkout_session'` and
+            `referenceId: $sessionId`. Both values reach the wire
+            through `GetCreditUsageQueryHandler`. **A stale "not yet"
+            retires a shipped capability and nothing detects it** — the
+            claim reads as caution, and a client that believes it stops
+            looking.
         created_at:
           type: string
           format: date-time
@@ -8528,6 +8686,9 @@ components:
           rule when product/Lambda confirms detection logic.)
 
         Per ticket [I25 `i5yCuSZc`](https://trello.com/c/i5yCuSZc).
+      # `WarningType` carries the NORMATIVE statement of the additive convention
+      # ("SDKs MUST treat the enum as additive"), so it is open by its own declaration.
+      x-enum-openness: open
       enum:
         - redundant_pre_encode_before_reencode_merge
 
@@ -8839,6 +9000,14 @@ components:
         429 response). **The per-tier numbers are deliberately not restated
         here** — read them from the source above.
 
+      # CLOSED, AND DELIBERATELY SO. The values are an ORDERED ladder driving upgrade
+      # prompts and tier comparisons — an unknown tier cannot be RANKED, so a consumer
+      # that silently accepted one would place it arbitrarily in that ordering.
+      # Failing on an unknown tier is the correct behaviour.
+      # SURVEY TRAP: this schema's description mentions "typed-open", but that
+      # sentence is about `AccountLimits.limits`, NOT about this enum. A keyword
+      # sweep reads it as open. It is not.
+      x-enum-openness: closed
       enum:
         - basic
         - free
@@ -9459,6 +9628,9 @@ components:
           server. Equivalent to `cancelled` for downstream
           accounting; distinct status so dashboards can surface the
           natural-expiry vs caller-initiated distinction.
+      # Declared "additive widening; V1 clients ignoring unknown enum values
+      # continue to function".
+      x-enum-openness: open
       enum:
         - pending
         - in_progress
@@ -9486,6 +9658,8 @@ components:
         - `in_progress`: At least one operation is running.
         - `completed`: All operations completed successfully.
         - `failed`: Job failed (at least one operation failed).
+      # Declared "additive widening".
+      x-enum-openness: open
       enum:
         - pending
         - waiting
@@ -9559,6 +9733,13 @@ components:
         V1 `watermark` operation removed at V2 cutover (I4-CONS) —
         replaced by `image_watermark` + `text_watermark` per ADR-0004
         §"Greenfield V2.0 cutover".
+      # CLOSED on BOTH sides, and the two sides declare DIFFERENT VALUE SETS on
+      # purpose — see schemas/cross-document-enums.yaml. openapi types the CREATE
+      # SURFACE; asyncapi types the PAYLOAD FIELD, which stays coarse because the
+      # thumbnail sub-types travel on the SNS message ATTRIBUTE, not in the payload.
+      # Closed because an unrecognised operation type cannot be dispatched or
+      # rendered; guessing is never right.
+      x-enum-openness: closed
       enum:
         - compress
         - thumbnail
@@ -9668,10 +9849,31 @@ components:
             use the multipart flow
             (`POST /api/uploads/multipart/initiate` → chunk PUTs →
             `POST /api/uploads/multipart/complete`). 10 MB /
-            10,000,000 bytes — chosen to match the ALB-fronted
-            single-request body cap. Raising this requires
-            infrastructure work (ALB swap or rearchitect; out of
-            scope for this ticket).
+            10,000,000 bytes — matching the **API Gateway HTTP API
+            payload quota** that fronts `api.<domain>`.
+
+            🔴 **THIS CANNOT BE INCREASED.** AWS documents the HTTP API
+            payload limit as 10 MB, "Can be increased: No". It is a
+            platform quota, not a tunable setting, so there is no
+            configuration change or support request that raises it —
+            only a different ingress would.
+
+            ⚠️ **This line previously said the cap matched an
+            "ALB-fronted single-request body cap" and that raising it
+            required an "ALB swap".** There is no ALB in front of
+            `/api/uploads`; the ALB module is instantiated only for
+            `stream.<domain>`. `compression_api` measured this on
+            2026-08-28.
+
+            ⇒ **The NUMBER was right and the RECORDED REASON was wrong,
+            which is the more dangerous way round**: a correct value
+            never triggers a review, and the rationale is what the next
+            person acts on. **And the error pointed the permissive
+            way** — it named a load-balancer setting somebody could
+            raise, where the real constraint is a non-adjustable quota.
+            Anyone revisiting this would have gone looking for a load
+            balancer that is not there, concluded the cap was tunable,
+            and planned work that cannot succeed.
         multipart_chunk_size:
           type: integer
           format: int64
@@ -10406,6 +10608,22 @@ components:
         The two forms are mutually exclusive (a request supplies exactly
         one — enforced by the `oneOf` below); they are NOT combined.
 
+        🔴 **THE FLAT FORM IS `planned`, AND THE SERVER DOES NOT ACCEPT IT
+        YET.** Send `jobs[]`. A flat request is rejected today, and the
+        rejection is currently a generic *"At least one job is required"*
+        rather than the `feature_not_available` (422) envelope this contract
+        requires for a `planned` shape — tracked against `compression_api`,
+        so do not read that generic message as a validation error in your
+        own payload.
+
+        ⚠️ **The equivalence stated above is the DESIGN, not observed
+        behaviour.** It describes what the server will do once the lowering
+        ships. It carried no availability marker from v2.64.0 until
+        2026-09-10, which per ADR-0001 §1.4 meant it read as `stable` — a GA
+        claim about a shape nothing accepts. Neither shipped SDK ever emitted
+        it: the TypeScript and PHP builders both always send `jobs[]`, so the
+        exposure was a hand-built payload and this document's own claim.
+
         Because the flat form lowers to ONE single-input job, its
         `operations[]` admits only **single-input** operation types. The
         multi-input / role operations (`merge`, `archive`,
@@ -10421,12 +10639,31 @@ components:
         otherwise behave exactly as for one explicit job.
       oneOf:
         - title: Explicit jobs form
+          # Stated rather than defaulted. This IS the served form, so `stable` is
+          # the value it would have inherited anyway — and that is exactly why it
+          # is written down: an unmarked branch and a deliberately-stable branch
+          # are indistinguishable on the page, which is how the flat form below
+          # spent three months claiming GA. Every form answers for itself.
+          x-availability: stable
           required: [jobs]
           not:
             anyOf:
               - required: [source]
               - required: [operations]
         - title: Flat single-job form
+          # 🔴 PLANNED — DECLARED, NOT SERVED. The server rejects this form
+          # today (`compression_api` 2.198.0 answers "At least one job is
+          # required"), so the equivalence stated in the description above is a
+          # DESIGN, not a behaviour. Hub ruling 2026-09-10, decision 192.
+          #
+          # ⚠️ It carried NO availability marker from v2.64.0 until now, which
+          # per ADR-0001 §1.4 means it defaulted to `stable` — the contract made
+          # a GA claim about a shape nothing accepts, for three months.
+          # `planned` is the honest value and the reason it is not simply
+          # deleted: the thin-alias design was ratified and codex-reviewed
+          # (`D0Gsri8V`), and deleting it would throw the design away to make a
+          # document true.
+          x-availability: planned
           required: [source, operations]
           not:
             anyOf:
@@ -11332,6 +11569,10 @@ components:
         value is that sentence. Emit it in preference to a
         measurement-asserting value whenever the metrics were absent,
         even if the class ultimately chosen happens to be correct.
+      # OPEN BY CITED CONVENTION, AND IT SAID SO NOWHERE. `WarningType` names it in
+      # the precedent set and this declaration was silent — so a generator reading
+      # only this schema could not learn it. That is the gap this marker closes.
+      x-enum-openness: open
       enum:
         - within_short_form_limits
         - input_size_exceeds_short_form
@@ -11757,6 +11998,11 @@ components:
           could not extract enough metadata to make a routing
           decision. Caller may proceed at their own risk; the
           workflow-create routing fallback applies.
+      # OPEN. `ok` is the only proceed value and everything else is a reason not to,
+      # so a consumer branches on `ok` rather than enumerating failures. A new
+      # failure reason is additive and an old consumer treats it correctly as
+      # not-ok.
+      x-enum-openness: open
       enum:
         - ok
         - corrupt
@@ -11773,6 +12019,11 @@ components:
         the file (e.g. free-tier caller probing a long-form clip,
         or any caller probing a `corrupt` / `unsupported_codec`
         file).
+      # CLOSED. It GATES AN AFFORDANCE: `blocked` is a hard stop and the others
+      # predict routing. An unrecognised class cannot be safely treated as routable —
+      # a consumer guessing would offer an upload the server will refuse. A new
+      # class must reach consumers deliberately.
+      x-enum-openness: closed
       enum:
         - short_form
         - long_form
@@ -14191,17 +14442,49 @@ components:
             once no consumer reads it.
 
             🔴 **KNOWN READER OF `environment`, RECORDED SO THE REMOVAL
-            CONDITION IS CHECKABLE RATHER THAN ASSUMED:**
-            `compression_api` reads it — `CapabilitiesSidecar::fromArray`
-            does `$raw['environment'] ?? null` and `OperationSchemaController`
-            emits it conditionally (reported by that session, 2026-08-25).
-
-            ⚠️ **THEIR ONLY TRIPWIRE IS A TEST THAT REMOVAL TURNS RED AND
-            DEPRECATION DOES NOT** — it asserts `environment === 'baseline'`
-            against the real vendored sidecar. So the null-coalesce means a
-            removal drops the field from `/api/operations/schema` **with no
-            error anywhere**, and the deprecation window generates no signal
-            they can act on.
+            CONDITION IS CHECKABLE RATHER THAN ASSUMED:** `compression_api`
+            reads it.
+
+            **AS MERGED ON THEIR `main` — api #662, merge commit `25801c8d`,
+            verified by ref 2026-08-28, not taken on report:**
+
+            - the read is **successor-preferred**: `capabilities_profile`
+              when provided, `environment` only as a fallback for a sidecar
+              predating the rename. `null` and `''` count as *not provided*
+              and fall back; a **present-but-wrong-type successor does NOT
+              fall back**, so a corrupt successor is never masked by the
+              predecessor's value.
+            - `DeprecatedEnvironmentKeyRemovalTriggerTest` fires on **four**
+              conditions — this contract **stops declaring** `environment`,
+              **loses `deprecated: true`**, **stops naming its successor**
+              via `x-superseded-by`, or **`availability.json` stops carrying
+              `environment`**. ⇒ **api goes red on the removal by itself**,
+              whichever artefact moves first.
+
+            🔴 **THE SIDECAR IS STILL A LEVER — ON THEIR BUILD, NOT ON THEIR
+            WIRE, AND THE TWO NOW POINT OPPOSITE WAYS.** api emits
+            `environment` as a **MIRROR of the resolved profile** rather than
+            as an independent read of the sidecar, so dropping it from
+            `availability.json` **no longer changes their wire output** — it
+            used to. It does now turn their **CI test suite** red, via
+            `testTheSidecarStillCarriesTheDeprecatedKeyItIsMirroredFrom`,
+            proved by mutation rather than by reading.
+            ⛔ **That is a PHPUnit alarm, NOT a build or deploy gate.** Their
+            production image installs without dev dependencies and neither
+            `ci-deploy.yml` nor `deploy-prod.yml` runs PHPUnit — `phpunit`
+            appears in `ci.yml` alone (verified against their `origin/main`).
+            ⇒ **An affected revision can BUILD AND DEPLOY, with the alarm
+            arriving separately.** The control gives detection, not
+            prevention, and the two must not be read as one.
+            ⚠️ **Under expand → migrate → contract the sidecar half fires
+            FIRST**, because the normal ordering drops the key there before
+            retiring the declaration here. ⇒ **Retiring the key quietly via
+            the sidecar ahead of this schema no longer half-works silently;
+            it fails loudly**, which is the outcome to want.
+            ⚠️ **That decoupling is the deliberate price of mirroring**, which
+            is what stops the successor drifting from the field it
+            supersedes; the four-condition control is what pays for it.
+            Recorded here rather than discovered later.
 
             ⇒ **REMOVAL IS A CO-LAND, NOT A CUT.** Tell every reader listed
             here BEFORE the removal ships, and remove a name from this list

package/operation-capabilities/operation-capabilities.json

@@ -51,7 +51,7 @@
       "accepts": [
         "audio"
       ],
-      "availability": "beta",
+      "availability": "stable",
       "input": {
         "max": 2,
         "min": 1,
@@ -638,7 +638,7 @@
       "accepts": [
         "video"
       ],
-      "availability": "beta",
+      "availability": "stable",
       "input": {
         "max": 2,
         "min": 2,
@@ -732,5 +732,5 @@
       "isAnimated": "maybe"
     }
   },
-  "schema_version": "2.198.0"
+  "schema_version": "2.201.0"
 }

package/operations/schemas/audio_to_video.yaml

@@ -2,7 +2,7 @@ schema_version: "2.0.0"
 
 operation:
   type: audio_to_video
-  availability: beta
+  availability: stable
   description: |
     Produce a video from an audio input plus an OPTIONAL still image
     overlay. Use cases: podcast → YouTube uploads, audio-only tracks

package/operations/schemas/video_watermark.yaml

@@ -2,7 +2,7 @@ schema_version: "2.0.0"
 
 operation:
   type: video_watermark
-  availability: beta
+  availability: stable
   description: |
     Apply an image overlay onto a base video
     asset. Multi-input role-based:
@@ -62,7 +62,7 @@ operation:
         - video/webm
       processing_class:
         short_form:
-          availability: beta
+          availability: stable
           constraints:
             max_input_duration: "PT5M"
             max_input_size_bytes: 524288000

package/package.json

@@ -1,8 +1,8 @@
 {
   "name": "@giveitsmaller/contracts",
-  "version": "0.68.0",
-  "description": "Generated contract types for GISL (Give It Smaller). Generated from compression_contracts spec v2.198.0.",
-  "gislContractsSpec": "v2.198.0",
+  "version": "0.70.0",
+  "description": "Generated contract types for GISL (Give It Smaller). Generated from compression_contracts spec v2.201.0.",
+  "gislContractsSpec": "v2.201.0",
   "license": "Apache-2.0",
   "homepage": "https://docs.giveitsmaller.com",
   "bugs": {