@openwop/spec-artifacts 2.46.11 → 2.46.13

package/CORPUS-STAMP.json

@@ -1,17 +1,17 @@
 {
   "_comment": "Provenance of @openwop/spec-artifacts (RFC 0168 §D.2). files: SHA-256 per file; the conformance suite compares the installed peer against dist/spec-artifacts.lock.json at start.",
   "package": "@openwop/spec-artifacts",
-  "version": "2.46.11",
-  "corpusTag": "v2.46.11",
+  "version": "2.46.13",
+  "corpusTag": "v2.46.13",
   "files": {
     "api/.redocly.lint-ignore.yaml": "bf5a8350b88a72fa43f59605ed8d903ed24b6cfccda5e45509c9f6ed9ee4e712",
     "api/asyncapi.yaml": "d5ecb9ee6114582be3b1f662c84bfac9ae96dae7bacb853e461168f70a8e1c7d",
     "api/grpc/openwop.proto": "c3e72bb17cba514ee98feb6434e6c9b6ea6795bfd086489ec69fd882dd1ad977",
     "api/openapi.yaml": "ac6de5303872664b78ff419c035915593e01599e02835e777a0123de1baeb398",
     "api/redocly.yaml": "b0604c89b2ca6d5076ec25725c539dad44a741a811fe524439ee6daef8baa09f",
-    "api/seams-v2.yaml": "7722f9cf3dcf9e4288bf1ad0087f347fbc1ee53ccd1f27ebde4b01cd9dfe2ba2",
-    "api/v2/asyncapi.yaml": "2faef7d29e10478bfcde17e63f34b436cedc8b8f6aecb44ca8895bd93ad44ebe",
-    "api/v2/openapi.yaml": "78c72788d422b216286a874a2900674cde0222a80cb92b29b9427d15a7d220e7",
+    "api/seams-v2.yaml": "eaec2c50979e9b55ac575d2be5c490a3048ae86613b1c018a59c1e8e0a26f24b",
+    "api/v2/asyncapi.yaml": "656c8c4031b8ebd96ad1bd536b64456f9eaf46249311e3332d540e4fbfcb63d7",
+    "api/v2/openapi.yaml": "473b40600965a068ab54e77e7efc09e9be4ed7eff670f015158b4b68aaad1e83",
     "api/v2/redocly.yaml": "1e66b60e6118ad11a823bb620678be464d99dfe50a40e3e6f93ec9429b88b34c",
     "schemas/README.md": "276fc1d51eb08e14e9127b9fd27761d7046ee33c12000f3a64a801322db8bf58",
     "schemas/a2a-task-state.schema.json": "75d5049dea8bd873ff0e7546f1c60c8d36c219bec7084264360be54a8be30ae1",
@@ -103,7 +103,7 @@
     "schemas/v2/a2ui-surface-delta-frame.schema.json": "6889c5da3fe7e76cb614038c0eea6774776fa640448558b258eb87fb50cb9b64",
     "schemas/v2/agent-deployment-transition.schema.json": "93a334ff042d51378c89fdd7134664271bf9592419bb5e3e00821ed06bc1904e",
     "schemas/v2/agent-deployment.schema.json": "d59b43fb62b522005ee22bc3bc53775a6777493e5e42d71812abc347625fa7fe",
-    "schemas/v2/agent-eval-suite.schema.json": "e80439bc6873a5d9a14122f81d16f7650104018d814bd316aafd3fc8b8225604",
+    "schemas/v2/agent-eval-suite.schema.json": "a49ad7a232ec781512375001f8d8c70ab073b69ca094f5251dbb30a0bbd972a7",
     "schemas/v2/agent-inventory-response.schema.json": "5f74c3da946dfc3114624567a989717748081ab947b4d2af3f49034d67f4306b",
     "schemas/v2/agent-manifest.schema.json": "004ff150e955deebb9e64273de5093f5144b9b3895b09953e65aa74b61140778",
     "schemas/v2/agent-org-chart.schema.json": "113b079e1dce95908467cbef7f83e076fdb82aab5b0678fc1a43e32bfc9cbadf",
@@ -117,7 +117,7 @@
     "schemas/v2/artifact.schema.json": "85b9dc284d39b99d2a66fa8cbf47f2dd3d8526d3bd6cc411ad315b372de38975",
     "schemas/v2/audit-verify-result.schema.json": "14af8e9283510f0a40e48a2059c929ee44e0fc9fa3cc66a95bfbe90b17c48297",
     "schemas/v2/budget-policy.schema.json": "1de68095db0beb5c7a4be629de26e79f62ebbe748a88d46f36e768edfd1f0d68",
-    "schemas/v2/capabilities.schema.json": "a717d34d98201cefe1769421c468d22e124d1a46881f60683cb6e103843dbc9a",
+    "schemas/v2/capabilities.schema.json": "87faea080e8043abd16cba19dece777a810731fc3e4455c96c02d0bfd0997d84",
     "schemas/v2/certification-bundle.schema.json": "5cfa7fb5247160333e33a58b60d42bcd4e5269d8391d6a112d301e14828892ed",
     "schemas/v2/channel-presence-payload.schema.json": "fc03df0c20cc12a25f7ee6a47503a4c5931d79fe3d5ff24815e8fd7c1bcb9a87",
     "schemas/v2/channel-written-payload.schema.json": "ccecff3c71a3275ad8db035ac0e04db25e6f8d23cf091ddbcd5a60d1be1cbb6b",
@@ -145,7 +145,7 @@
     "schemas/v2/envelopes/schema.response.schema.json": "d049a081267c06c28879c331b147f3ed821b14ccfd93005c67aff64db0a8c68e",
     "schemas/v2/envelopes/ui.a2ui-surface.schema.json": "e92a2548ba65454c9e0a34f72ea6768d18f59e61fa5da7668d0b2d5baddaddc8",
     "schemas/v2/error-envelope.schema.json": "8ddb1d80330dfef17572a74dfe3a53e1a06411e4b0af6a6f1dd6e7e7461da987",
-    "schemas/v2/eval-summary.schema.json": "e670fad6c4c487f80edf3832ccd427b534b2066fbbc1cc2c158e205d6e7e3f6b",
+    "schemas/v2/eval-summary.schema.json": "d07f4d116c5410997a81d0bc5387545fb61933a6333c360a26ab8a71c749336b",
     "schemas/v2/export-bundle.schema.json": "bd28905a7729437ab7f2ad8bb33e13a371b2bf5605a7eb2962a0d246de312d69",
     "schemas/v2/form-content-pack-manifest.schema.json": "a1942a124e250a6e9fbdb6b811c5a125e78fbcbcaa46dfc71d0636f24710a339",
     "schemas/v2/frontend-plugin-manifest.schema.json": "b279da288d542906a72b8cb4e872e001de632eaff539e0259b9cce287f82bf8f",
@@ -208,14 +208,14 @@
     "schemas/workspace-file.schema.json": "464de85c2a068243084ee9c1d969bc7cd5d8f7948574e58450d6493c38a0e1e4",
     "spec/v1/alias-detectors.json": "22444c22bb88c568339ef512553f9fc38146e22b566d11ab8608f655d688d19c",
     "spec/v1/capability-declaration-classes.json": "e7729aed5c4b4e1dd02abab0530f14cc95f5d4070fe51fb139e7f5cccefa00c6",
-    "spec/v1/core-standard-manifest.json": "2a845db8a673c2f5c95ff752fc7765fb72779abf589595e7bf9989f36216ea39",
+    "spec/v1/core-standard-manifest.json": "1070771f6480ead748c5bf02f0cfbc397ce0009dc4dd237c58d1178360c91179",
     "spec/v1/deprecations.json": "e723bc11d7272c71789ad1303bb7f21809a8a024abf57594f9d5c35dc199ece9",
     "spec/v1/deprecations.schema.json": "3e393c405d2a41b467d8c5e3c468549078df1ce6a6d2588fc488097b95d9b55b",
     "spec/v1/end-of-support.json": "35f0bf3b70fe78b16e2a2fd81a153fc27e703f6c16376c6fe7cdf1f4f2c4e6d4",
     "spec/v1/event-codemap.json": "3da60d884157793a360da532a9fcbbfb5285636db325a74cec94b34622186d97",
     "spec/v1/event-codemap.schema.json": "d05933b2e88103aff51a2f774df97b9bda065e0fc88dbd9b114cc47a32189174",
     "spec/v1/extensions.json": "79a60754aa16cbdbdb604f8e5c00038af535a1690a4d07bd5ff7c38f0256a7bc",
-    "spec/v1/gaps.json": "3bd864a64f6c9c69a3bfc4178d9c7df8dd625b682c1dbde1656b89a239c5a6f9",
+    "spec/v1/gaps.json": "ced525510fbb3405dae0cdff4f0943a6d2aaa1bbff39312faac2679a55952116",
     "spec/v1/gaps.schema.json": "8fd83259f556553c9df0f53e7a82ca8c2a4771d471197f69b4e2ae0ceeacfd66",
     "spec/v1/migrations.json": "4c50760be70cb644d02427bcaf5749606700edf4513f423903cf3b1d12f5c300",
     "spec/v1/migrations.schema.json": "886779aa6c22e646db097f5df210adb018a4dd14a7b815465a18c8a7056c8f72",
@@ -228,7 +228,7 @@
     "spec/v2/core/connection-packs.md": "c80f95bd6a03698f35c72cdb2b46bd973f49203bbd786c26760e35fa86e6f23d",
     "spec/v2/core/conversation.md": "8d6abedf6af408a74577753bfca24719d318e2632e10b45ff5d63bdcf31a8aaa",
     "spec/v2/core/errors.md": "e48838701b86bbfe0ffee77b18a8ffc4e5f04724b4318507d9771da653f8da0f",
-    "spec/v2/core/events.md": "370621629c2439747f2e4ddab4780eea4345354f79e5098739d66823873f3105",
+    "spec/v2/core/events.md": "c3dcece6f1b1aa43e1e01c5ece23b2df075a505d5f6c2f37de0801aeaa486b62",
     "spec/v2/core/execution.md": "4218dbb397534940f83ca5d46561f02bf8f3fdf68d2ae7fefe7f2eb2ed771103",
     "spec/v2/core/form-content-packs.md": "02931227899676fd4581a27017be2a6c4953904717be9b81c29d12f87ab3917d",
     "spec/v2/core/headers.md": "f69fe81e4a3ad21879bb85948660c16860eb05bf152928c31aa4898f6d7d1887",
@@ -254,7 +254,7 @@
     "spec/v2/core/workflow-chain-packs.md": "fdce7a8477e269950a2bac05ee6eddbd66eb0b670b464d00054f7196ec05d29c",
     "spec/v2/corrections.json": "3a8d93e33865b692d26a3b6bfd3819766e206b2e0c2a69669baa4bf833d05a49",
     "spec/v2/corrections.schema.json": "46bbffb954d13bd9995aebc912ceeb2544f7064331abf25fd6ab85c81a5cec94",
-    "spec/v2/declaration.json": "773af65df59ce7bc2b3ab17093e23ac8083a8faaeff643302fe30198deae933b",
+    "spec/v2/declaration.json": "de8558d76a03d07101c5e26a7311d3e90b449f065c9ba418f4abd9c91864a58b",
     "spec/v2/declaration.schema.json": "ef951633e4070899f64828f5cf1b5f4870270a0360f81dbb49b5b4d6cd407550",
     "spec/v2/eos-override.json": "4c72d2f6a658943b01c1de1643cf294642091933af7832f38ee5495df684cb2d",
     "spec/v2/errors.json": "1f0e29455edbd22e28f67f221766209c3c55ac4987beb77993b5702dd7cae131",
@@ -281,7 +281,7 @@
     "spec/v2/ext/sandbox-runtime-notes/README.md": "0d1290fb1e20b2e35c0b15051541ab6546e4e2a6f693b4f912cb3f7d6a65e92c",
     "spec/v2/ext/webResearch/README.md": "b675e6f6646128d781b661f754c2393eed86e5935d16c3a22df582a5ce025b9b",
     "spec/v2/facets/a2a.schema.json": "b68bf5e29e455e46c2ad7cfcccf5963730ec2310ff9db81884550d8df30fd6ef",
-    "spec/v2/facets/agents.schema.json": "62a979d23219ca684bc16f91f81a4c92c2cb56e0f084d2e4f694c03e0c7125df",
+    "spec/v2/facets/agents.schema.json": "88fe144572c25e1b09b99fd83e2ad12b2611d4d7366b5d53c3a50d0310faad93",
     "spec/v2/facets/aiProviders.schema.json": "4eb911c2403b8f618325194855a1326680fbda2549dfbfe86a70dd2743f8fb12",
     "spec/v2/facets/auditLogIntegrity.schema.json": "9a7f81320c8f940c6f4129bc7dd895a4063231c5458a717a08c2cbf9bb886c46",
     "spec/v2/facets/auth.schema.json": "7d98bc81a0f7d307460d039eebfebd21388091bfe1d538762583940f94c0bd2a",
@@ -290,6 +290,7 @@
     "spec/v2/facets/heartbeat.schema.json": "81d7fa1895a8b0241bd07cf00a7e7905bef0fd14eb3427325261c62eee5746a6",
     "spec/v2/facets/hostEvents.schema.json": "973ff452cb91356b4c625d0b06c8d2ef335e18183de09e1afde8acead22290b3",
     "spec/v2/facets/interrupt.schema.json": "2110ef32c3f880bf706475ef4cf8ddf97ff06871b8e6003a5edbcf4439f1fb70",
+    "spec/v2/facets/limits.schema.json": "218fcbd34dd2cdf7548c29c8c76e1a6aea2b7d7df1ba32fbe201bd40917bc66f",
     "spec/v2/facets/mcp.schema.json": "7060657dfcff170f079beed160e3db7c1d3cc4410cd3cf64863baf6459f781cd",
     "spec/v2/facets/nondeterminismPolicy.schema.json": "e9dfaa32e0778a1a837fdad3b8a0673fccda9579ab08c8bfe9777362dd15e989",
     "spec/v2/facets/oauth.schema.json": "3a0cbe249d3bd5a5982349ae7ba2534776cc958e4a57e0fbeabf2bdab6e99e69",
@@ -313,9 +314,9 @@
     "spec/v2/path-manifest.json": "7b5679a4dad98c80d855361d88a98c8a4e045af15e310862a05b285b815dc9fc",
     "spec/v2/peer-dependency-aliases.json": "d10299280abee08258502925bc327293ee413e0108cd6e6ec75ff6110653308d",
     "spec/v2/profiles.json": "b382b261c7844798575f7c1b37c934f0da19846016a2337d883d4f5ef6e05479",
-    "spec/v2/release.json": "203adbfd38c222e739f8db654cf2bb453d61dcbdb74702949d933fa3dbad9baf",
+    "spec/v2/release.json": "e8fe4796dd85264ee4877b203da962d8c593c996e428289b165827cd86bfd2cc",
     "spec/v2/retention-floors.json": "eaf3722d95c79947af1d4269ef85117e126518c588cfcf1a2b21b97269f51624",
-    "spec/v2/surface-baseline.json": "a566e5a039f75a076bde41640f4ecbea50fbae41bf4581aff69c2dad14b24b6a"
+    "spec/v2/surface-baseline.json": "094912462a9b1bdd8bc3d7c9f10b4dc342d0db41b2802c3a1c60d8a8b27d2860"
   },
-  "corpusCommit": "398b842f0e19e8de6cd6b6c706b51bedcff77356"
+  "corpusCommit": "234f6c57b1e01099827dcfabeaad542c2604b17a"
 }

package/api/seams-v2.yaml

@@ -1,7 +1,7 @@
 openapi: 3.1.0
 info:
   title: OpenWOP conformance seams profile (openwop-conformance-seams-v2)
-  version: 2.46.11
+  version: 2.46.13
   description: 'The OpenWOP conformance seams profile (`openwop-conformance-seams-v2`): test-only operations in their own
     path space, `/conformance/seams/…`. The profile is versioned and validated against the canonical v2 schemas with no tolerance
     path. It is not a capability: a host advertises the profile, never a `testSeams` flag.

package/api/v2/asyncapi.yaml

@@ -1,7 +1,7 @@
 asyncapi: 3.1.0
 info:
   title: OpenWOP v2 event streams
-  version: 2.46.11
+  version: 2.46.13
   description: 'OpenWOP v2 event streams over SSE. The run-events channel''s address is the OpenAPI path key, and the server
     pathname is empty (bare origin). `streamMode` is a stream mode from the closed set, or a comma-separated combination of
     them; `values` never combines. `hostEvents` has a default address, `/host/events`; a host MAY declare another under `heartbeat.deliveryChannel`.
@@ -56,6 +56,10 @@ operations:
               type: string
               pattern: ^(values|(updates|messages|debug)(,(updates|messages|debug))*)$
               default: updates
+            resumable:
+              type: boolean
+              default: false
+              description: RFC 0247 opt-in to a bounded close (events.md §Bounded close).
   subscribeHostEvents:
     action: receive
     channel:
@@ -64,6 +68,14 @@ operations:
     bindings:
       http:
         method: GET
+        query:
+          type: object
+          additionalProperties: false
+          properties:
+            resumable:
+              type: boolean
+              default: false
+              description: RFC 0247 opt-in to a bounded close (events.md §Bounded close).
 components:
   securitySchemes:
     ApiKeyAuth:

package/api/v2/openapi.yaml

@@ -1,7 +1,7 @@
 openapi: 3.1.0
 info:
   title: OpenWOP v2 API
-  version: 2.46.11
+  version: 2.46.13
   summary: REST surface for declaring, executing, suspending, resuming, and observing multi-step workflows.
   description: 'The OpenWOP v2 REST API. Paths are unversioned and served at the bare origin; a client selects a version with
     the `OpenWOP-Version` header from the host''s `protocolVersions[]`. Conformance seams and test-mode operations are not
@@ -511,6 +511,15 @@ paths:
           trigger fires (terminal events, suspensions, connection close), and emits a single SSE event with `event: batch`
           and `data:` as a JSON array of `RunEventDoc`. Range 0..5000; `0` means no buffering. See `stream-modes.md` §Aggregation
           hint.'
+      - in: query
+        name: resumable
+        schema:
+          type: boolean
+          default: false
+        description: RFC 0247. `true` says this consumer resumes with `Last-Event-ID` when the stream closes without the terminal
+          event, so a host advertising `limits.maxStreamAgeSec` MAY end it early (a bounded close, never terminal). Absent
+          or `false`, the host MUST hold a live stream open (events.md §Resuming with `Last-Event-ID`). A consumer that sends
+          it MUST resume with `resumable=true` and `Last-Event-ID`.
       - in: header
         name: Last-Event-ID
         schema:
@@ -3550,6 +3559,15 @@ paths:
         '401':
           $ref: '#/components/responses/Unauthenticated'
       parameters:
+      - in: query
+        name: resumable
+        schema:
+          type: boolean
+          default: false
+        description: RFC 0247. `true` says this consumer resumes with `Last-Event-ID` when the stream closes without the terminal
+          event, so a host advertising `limits.maxStreamAgeSec` MAY end it early (a bounded close, never terminal). Absent
+          or `false`, the host MUST hold a live stream open (events.md §Resuming with `Last-Event-ID`). A consumer that sends
+          it MUST resume with `resumable=true` and `Last-Event-ID`.
       - $ref: '#/components/parameters/OpenWOPVersion'
       - $ref: '#/components/parameters/OpenWOPClientVersion'
       security:

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@openwop/spec-artifacts",
-  "version": "2.46.11",
+  "version": "2.46.13",
   "description": "The OpenWOP machine-readable contract: api/ (OpenAPI, AsyncAPI, the seams profile), schemas/ (v1 and v2 JSON Schemas), the spec/v1 and spec/v2 registries (errors, event codemap, declaration, deprecations, migrations, gaps) and CORPUS-STAMP.json — published from the openwop/openwop corpus tag; the peer dependency @openwop/openwop-conformance digest-checks at start (RFC 0168 §D.2).",
   "license": "Apache-2.0",
   "repository": {

package/schemas/v2/agent-eval-suite.schema.json

@@ -105,7 +105,7 @@
             "mean"
           ],
           "default": "all",
-          "description": "A task passes when: `all`, every trial passed; `any`, at least one did; `mean`, the mean trial score >= `thresholds.passScore`."
+          "description": "A task passes when: `all`, every trial passed; `any`, at least one did; `mean`, the task gating score >= `thresholds.passScore` (RFC 0246 §B: the mean over trials of the trial score, with 0 for a trial that failed its trajectory or failed closed; `meanScore` itself is unchanged)."
         }
       }
     },
@@ -297,6 +297,11 @@
                         "subset"
                       ],
                       "description": "RFC 0244 §A. How `args` is compared with the call's `inputs`. `ignore` (default): not compared. `exact`: `inputs` JSON-deep-equals `args`. `subset`: every key path in `args` is present in `inputs` with a JSON-deep-equal value. The host compares the call as executed, before any wire redaction."
+                    },
+                    "isError": {
+                      "type": "boolean",
+                      "default": false,
+                      "description": "RFC 0245 §A. MAY. `true`: the recorded call failed, and `response` is the error payload. A host serving this entry MUST return it to the agent as a tool error and emit `agent.tool-returned` with `error` and no `outcome`; it is not a fixture miss and is never retried live. Requires `agents.evalSuite.fixtureErrors`."
                     }
                   }
                 }

package/schemas/v2/capabilities.schema.json

@@ -561,6 +561,11 @@
           "type": "integer",
           "minimum": 1,
           "description": "RFC 0094 §H. Maximum REST request body size (bytes) the host accepts. Optional v1 field per `capabilities.md` §3 (previously documented as reserved — the closed `limits` object made advertising it a schema-validation failure). Hosts that advertise it MUST enforce it."
+        },
+        "maxStreamAgeSec": {
+          "type": "integer",
+          "minimum": 15,
+          "description": "RFC 0247. The longest, in seconds, this host holds a live `streamRunEvents` or `streamHostEvents` stream that the consumer opened with `resumable=true`. Only for such a stream, a host advertising this facet MAY end it before the run's terminal event (a bounded close) once it has been open 15 seconds, and MUST end it by this age. It MUST end at a frame boundary, MUST flush any pending batch first, and SHOULD first send an SSE `retry:` field (milliseconds). A host MUST NOT end an opted-in stream early unless it advertises this facet, and MUST NOT end a stream opened without `resumable=true` early whatever it advertises (events.md §Resuming with `Last-Event-ID`). A bounded close is never a terminal signal; only the terminal event is. A consumer that sent `resumable=true` MUST treat any close without the terminal event (a bounded close, a network drop, an intermediary's timeout) as non-terminal, and MUST reconnect with `resumable=true` and `Last-Event-ID` set to the last `id:` it received (no header if it received none). It SHOULD wait the `retry:` delay, or its own default of about one second, and SHOULD back off when reconnects fail or close without a frame. A consumer MUST handle such a close whether or not this facet is advertised. On `hostEvents` the consumer resumes from the last durable event's `id:`; ephemeral events missed in the gap are not redelivered (events.md §Host events)."
         }
       },
       "allOf": [
@@ -599,7 +604,7 @@
         }
       ],
       "description": "core/capabilities.md#limits — witness: witnessable-gated; maturity stable/none",
-      "x-openwop-seeded-from": "v1"
+      "x-openwop-facets-from": "spec/v2/facets/limits.schema.json"
     },
     "envelopeStrictness": {
       "type": "object",
@@ -3328,6 +3333,10 @@
               "type": "boolean",
               "description": "RFC 0244 §D. `true`: the host honours `toolResponses[].args` / `argsMatch` (first unconsumed match) and `fixtures.onMiss`, and reports `fixtureMissCount`."
             },
+            "fixtureErrors": {
+              "type": "boolean",
+              "description": "RFC 0245 §B/§C. `true`: a `toolResponses[]` entry with `isError: true` is returned to the agent as a tool error and recorded as `agent.tool-returned` with `error` and no `outcome` (plus `status: \"error\"` under tool hooks); it is not a fixture miss and is never retried live. A suite carrying `isError: true` on a host that does not advertise this MUST be refused with `400 validation_error`."
+            },
             "regressionVerdict": {
               "type": "boolean",
               "description": "RFC 0244 §F. `true`: in `regression` mode the host reports `regression.verdict`, `method`, `alpha` and `pairedTaskCount`, pairing tasks by `taskId`."

package/schemas/v2/eval-summary.schema.json

@@ -43,9 +43,15 @@
       "maximum": 1,
       "description": "The suite-level score (0.0\u20131.0): the aggregation (host-defined, typically the mean) of per-task scores."
     },
+    "gatedAggregateScore": {
+      "type": "number",
+      "minimum": 0,
+      "maximum": 1,
+      "description": "RFC 0246 \u00a7A/\u00a7D. The aggregation used for `aggregateScore`, applied to the per-task `gatedScore`s instead of the task scores; suite `passed` reads it. MUST be present whenever any task has `expected.trajectory`. `aggregateScore` is unchanged (RFC 0246 \u00a7C)."
+    },
     "passed": {
       "type": "boolean",
-      "description": "Whether the run cleared the suite's `thresholds` (RFC 0081 \u00a7A) \u2014 `aggregateScore >= passScore` AND, when declared, `totalCostUsd <= maxCostUsd` AND the p95 latency bar. The load-bearing flag an RFC 0082 deployment gate may require (`requiredEval`)."
+      "description": "Whether the run cleared the suite's `thresholds` (RFC 0081 \u00a7A): the gating aggregate (`gatedAggregateScore`; `aggregateScore` when no task can gate) `>= passScore` AND, when declared, `totalCostUsd <= maxCostUsd` AND the p95 latency bar. RFC 0246 \u00a7B: a trial that failed its trajectory contributes 0 to the gating aggregate, so `passed` can be false while `aggregateScore >= passScore`. The load-bearing flag an RFC 0082 deployment gate may require (`requiredEval`)."
     },
     "taskCount": {
       "type": "integer",
@@ -85,6 +91,12 @@
             "maximum": 1,
             "description": "Task score (0.0\u20131.0)."
           },
+          "gatedScore": {
+            "type": "number",
+            "minimum": 0,
+            "maximum": 1,
+            "description": "RFC 0246 \u00a7A/\u00a7D. The task gating score: the mean over its trials of the trial score, with 0 for a trial that failed its trajectory or failed closed on a fixture miss. MUST be present on a task with `expected.trajectory` (equal to `score` when every trial matched); MAY be present on another task only when it differs from `score`. Under `passRule: \"mean\"` the task `passed` reads it; `score` is unchanged (RFC 0246 \u00a7C)."
+          },
           "passed": {
             "type": "boolean",
             "description": "Whether this task individually met its bar."

package/spec/v1/core-standard-manifest.json

@@ -4,7 +4,7 @@
   "provenance": {
     "suite": {
       "package": "@openwop/openwop-conformance",
-      "version": "2.46.11"
+      "version": "2.46.13"
     },
     "note": "Derived from the corpus at generation time. Regenerate with --write; verify with --check."
   },
@@ -426,5 +426,5 @@
       "$id": "https://openwop.dev/spec/v1/workspace-file.schema.json"
     }
   ],
-  "digest": "2ba970e2bc387bf1dc7fe1fd9caa8dcac67f461223da82112286d73971154973"
+  "digest": "a29cfbbca5f88e3db5efdf3358822f67ec34b3c9d3d6d6a7fc8f56f66262b856"
 }

package/spec/v1/gaps.json

@@ -1,7 +1,7 @@
 {
   "$schema": "./gaps.schema.json",
   "version": 1,
-  "updated": "2026-10-15",
+  "updated": "2026-10-16",
   "witnessClasses": [
     "witnessable-unaided",
     "witnessable-gated",
@@ -14717,6 +14717,142 @@
         }
       ]
     },
+    {
+      "id": "openwop.gap.0245.1",
+      "rfc": "0245",
+      "local": "G1",
+      "section": "§B, §Falsifiability",
+      "surface": "\"The agent receives the entry as a tool error\" cannot be caused from outside: the scripted conformance agent's results are scripted, so what it saw and which branch it took are not observable.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "transferred",
+      "target": "TODO.md",
+      "sources": [
+        {
+          "file": "RFCS/registers/0245-eval-fixtures-replay-tool-errors.gaps.md",
+          "token": "G1"
+        }
+      ]
+    },
+    {
+      "id": "openwop.gap.0245.2",
+      "rfc": "0245",
+      "local": "G2",
+      "section": "UQ 1",
+      "surface": "No registered error code for a failed tool execution, so a portable suite cannot assert `error.code`.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "transferred",
+      "target": "TODO.md",
+      "sources": [
+        {
+          "file": "RFCS/registers/0245-eval-fixtures-replay-tool-errors.gaps.md",
+          "token": "G2"
+        }
+      ]
+    },
+    {
+      "id": "openwop.gap.0245.3",
+      "rfc": "0245",
+      "local": "G3",
+      "section": "UQ 2",
+      "surface": "For a host-native (non-MCP) tool, the agent-facing error shape is host-defined.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "transferred",
+      "target": "TODO.md",
+      "sources": [
+        {
+          "file": "RFCS/registers/0245-eval-fixtures-replay-tool-errors.gaps.md",
+          "token": "G3"
+        }
+      ]
+    },
+    {
+      "id": "openwop.gap.0246.1",
+      "rfc": "0246",
+      "local": "G1",
+      "section": "§Acceptance",
+      "surface": "`COMPATIBILITY.md` §3 requires a `### Security` CHANGELOG entry citing an advisory ID per `SECURITY.md`; no advisory exists yet.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "closed",
+      "target": null,
+      "sources": [
+        {
+          "file": "RFCS/registers/0246-trajectory-failure-gates-the-pass.gaps.md",
+          "token": "G1"
+        }
+      ]
+    },
+    {
+      "id": "openwop.gap.0246.2",
+      "rfc": "0246",
+      "local": "G2",
+      "section": "UQ 1",
+      "surface": "Whether `gatedScore` is always present on a trajectory task, or only when it differs.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "closed",
+      "target": null,
+      "sources": [
+        {
+          "file": "RFCS/registers/0246-trajectory-failure-gates-the-pass.gaps.md",
+          "token": "G2"
+        }
+      ]
+    },
+    {
+      "id": "openwop.gap.0246.3",
+      "rfc": "0246",
+      "local": "G3",
+      "section": "§Compatibility",
+      "surface": "MyndHyve (build `73befc109`) implements RFC 0244 and decides both the `mean` task pass and the suite `passed` on output scores, so it is in the harm set; the window waiver is conditional on its fix.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "transferred",
+      "target": "TODO.md",
+      "sources": [
+        {
+          "file": "RFCS/registers/0246-trajectory-failure-gates-the-pass.gaps.md",
+          "token": "G3"
+        }
+      ]
+    },
+    {
+      "id": "openwop.gap.0247.1",
+      "rfc": "0247",
+      "local": "G1",
+      "section": "§B, §Falsifiability",
+      "surface": "No suite leg observes a bounded close: it needs an opted-in stream held open on a live run past the advertised age, longer than the suite's stream waits budget for.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "transferred",
+      "target": "TODO.md",
+      "sources": [
+        {
+          "file": "RFCS/registers/0247-bounded-stream-lifetime.gaps.md",
+          "token": "G1"
+        }
+      ]
+    },
+    {
+      "id": "openwop.gap.0247.2",
+      "rfc": "0247",
+      "local": "G2",
+      "section": "UQ 2",
+      "surface": "A host behind a short front end closes sooner on that hop than its advertised `maxStreamAgeSec`; a consumer cannot learn the hop's bound.",
+      "witness": "unclassified",
+      "requirementId": null,
+      "disposition": "transferred",
+      "target": "TODO.md",
+      "sources": [
+        {
+          "file": "RFCS/registers/0247-bounded-stream-lifetime.gaps.md",
+          "token": "G2"
+        }
+      ]
+    },
     {
       "id": "openwop.gap.spec.a2a-integration.g1",
       "rfc": "0152",

package/spec/v2/core/events.md

@@ -1,7 +1,7 @@
 # Events
 
 > **Status: Stable.**
-> **Normative home:** `heartbeat`, `hostEvents`, `envelopeContracts`, `envelopes`, `feedback`, `providerUsage`, `supportedEnvelopes`, `schemaVersions`, `envelopeStrictness`.
+> **Normative home:** `heartbeat`, `hostEvents`, `envelopeContracts`, `envelopes`, `feedback`, `providerUsage`, `supportedEnvelopes`, `schemaVersions`, `envelopeStrictness`, `limits`.
 
 ## Why this exists
 
@@ -100,7 +100,7 @@ Each frame carries `id:` (the `sequence`), `event:` (the v2 `type`, in a single
 - `batch` — `bufferMs`; `data:` is an array of `RunEventDoc`.
 - `ai.message.chunk` — `messages` mode; `data:` is the `outputChunk` payload, persisted as type `output.chunk`.
 
-A host MUST set `Content-Type: text/event-stream`, MUST emit a keep-alive comment at least every 30 seconds, and MUST close the connection after the run's terminal event (`run.completed`, `run.failed`, `run.cancelled`).
+A host MUST set `Content-Type: text/event-stream`, MUST emit a keep-alive comment at least every 30 seconds, and MUST close the connection after the run's terminal event.
 
 #### Resuming with `Last-Event-ID`
 
@@ -112,6 +112,10 @@ A host MUST set `Content-Type: text/event-stream`, MUST emit a keep-alive commen
 - The header is evaluated only after the caller is authorized to read the run. For a run the caller cannot read, the response MUST be the one the host gives without the header.
 - In `values` mode, resumption MUST emit a `state.snapshot` first.
 
+#### Bounded close
+
+With `resumable=true`, a host advertising `limits.maxStreamAgeSec` MAY end a live stream early, per that facet. That close is not terminal; the consumer MUST resume with `Last-Event-ID`.
+
 #### Batching and subscribers
 
 With `bufferMs` (0..5000) the host accumulates events into one `event: batch` frame whose `data:` is an array of `RunEventDoc`.
@@ -171,7 +175,7 @@ The response is `{ runId, events, lastSequence, status, isTerminal }` (closed):
 A run's log MUST contain exactly one terminal run event: `run.completed`, `run.failed` or `run.cancelled`.
 
 - After it, the log MUST NOT contain another terminal run event, `run.started`, `run.resumed`, `run.resume-started`, `run.paused`, `run.restored-from-snapshot`, any `node.*` event or any `interrupt.*` event.
-- `compensation.*` events and `run.dead-lettered` MAY follow it (a compensating host unwinds after a cancelled parent). Vendor-prefixed types are unconstrained.
+- `compensation.*` events and `run.dead-lettered` MAY follow it. Vendor-prefixed types are unconstrained.
 - A host that receives work for a run whose terminal event is recorded (a duplicate delivery, a late worker) MUST NOT append forward-execution events for it, SHOULD record the refusal in its operational log, and MUST NOT surface the refusal as a run event.
 
 ## Era-2 logs
@@ -189,7 +193,7 @@ An `eventLogSchemaVersion` of `2` means v1-written. Every reader, diff included,
 
 `supportedEnvelopes.kinds` is the catalog. A host advertising it MUST refuse an emitted `type` that is neither universal nor a member, with `unknown_envelope_kind`.
 
-An absent `kinds` is not an empty catalog and is not an unrestricted one: a host advertising `supportedEnvelopes` without it has made no catalog claim, and an engine MUST refuse every non-universal kind rather than admit it unchecked.
+An absent `kinds` is not an empty catalog and is not an unrestricted one: an engine MUST refuse every non-universal kind.
 
 ### `schemaVersions`
 
@@ -223,7 +227,7 @@ An absent `kinds` is not an empty catalog and is not an unrestricted one: a host
 A host advertising `providerUsage` MUST emit exactly one `provider.usage` per LLM provider invocation, before that node's `node.completed`. A host that does not advertise it omits the event.
 
 - `inputTokens` and `outputTokens` MUST replay identically; `costEstimateUsd` MAY be omitted on replay.
-- **`providerUsage.costEstimates`** advertises that the host stamps a derived cost on the event. That figure is an estimate from the host's own rate table; a consumer MUST NOT treat it as a billed amount.
+- **`providerUsage.costEstimates`** advertises that the host stamps a derived cost on the event. It is an estimate from the host's rate table; a consumer MUST NOT treat it as a billed amount.
 - The payload MUST NOT carry credential refs, hashed credential identifiers, or prompt or response text (invariant `provider-usage-no-credential-leak`).
 - `providerUsage.currency` is the ISO 4217 currency of `costEstimateUsd`; absent means USD.
 
@@ -253,4 +257,4 @@ Under `reliability.completion.distinguishesTruncation`, an envelope is complete
 - On exhaustion the host MUST emit `envelope.retry-exhausted` with that `finalReason` and `cap.breached` `kind: "schema"`, and fail the node with `envelope_truncation_unrecoverable` or `envelope_invalid`.
 - Lenient-parse recovery consumes no retry. It emits `envelope.recovery-applied` once per recovery, carrying only the path and an optional byte offset.
 
-*Sources: RFCs 0026, 0030, 0032, 0033, 0056, 0060, 0171, 0172, 0176, 0185, 0194, 0213, 0236.*
+*Sources: RFCs 0026, 0030, 0032, 0033, 0056, 0060, 0171, 0172, 0176, 0185, 0194, 0213, 0236, 0247.*

package/spec/v2/declaration.json

@@ -206,7 +206,7 @@
     {
       "key": "limits",
       "kind": "family",
-      "normativeText": ["spec/v2/core/runs.md"],
+      "normativeText": ["spec/v2/core/runs.md", "spec/v2/core/events.md"],
       "anchor": "core",
       "section": "core/capabilities.md#limits",
       "peerDependencyId": "limits",
@@ -224,7 +224,8 @@
         "maxLoopIterations",
         "maxBudgetTokens",
         "maxBudgetCostUsd",
-        "maxRequestBodyBytes"
+        "maxRequestBodyBytes",
+        "maxStreamAgeSec"
       ],
       "floorScenarios": [],
       "requirementIds": [],

package/spec/v2/facets/agents.schema.json

@@ -184,6 +184,10 @@
           "type": "boolean",
           "description": "RFC 0244 §D. `true`: the host honours `toolResponses[].args` / `argsMatch` (first unconsumed match) and `fixtures.onMiss`, and reports `fixtureMissCount`."
         },
+        "fixtureErrors": {
+          "type": "boolean",
+          "description": "RFC 0245 §B/§C. `true`: a `toolResponses[]` entry with `isError: true` is returned to the agent as a tool error and recorded as `agent.tool-returned` with `error` and no `outcome` (plus `status: \"error\"` under tool hooks); it is not a fixture miss and is never retried live. A suite carrying `isError: true` on a host that does not advertise this MUST be refused with `400 validation_error`."
+        },
         "regressionVerdict": {
           "type": "boolean",
           "description": "RFC 0244 §F. `true`: in `regression` mode the host reports `regression.verdict`, `method`, `alpha` and `pairedTaskCount`, pairing tasks by `taskId`."

package/spec/v2/facets/limits.schema.json

@@ -0,0 +1,60 @@
+{
+  "$comment": "RFC 0247 — the limits facets, hand-decided from the v1 seed with the nine seeded shapes unchanged, plus maxStreamAgeSec (bounded close of an opted-in event stream; events.md §Bounded close).",
+  "properties": {
+    "clarificationRounds": {
+      "type": "integer",
+      "minimum": 0,
+      "description": "Maximum total clarification envelopes per task before the engine forces the LLM to proceed."
+    },
+    "schemaRounds": {
+      "type": "integer",
+      "minimum": 0,
+      "description": "Maximum total schema-validation rounds per envelope before the node fails."
+    },
+    "envelopesPerTurn": {
+      "type": "integer",
+      "minimum": 0,
+      "description": "Maximum total envelopes the LLM may emit in a single chat turn."
+    },
+    "maxNodeExecutions": {
+      "type": "integer",
+      "minimum": 1,
+      "description": "Engine-side ceiling on total node executions per run. Acts as the upper bound for `RunOptions.configurable.recursionLimit`. Optional in v1; hosts that advertise it MUST enforce it."
+    },
+    "maxRunDurationMs": {
+      "type": "integer",
+      "minimum": 1000,
+      "description": "RFC 0058. Engine-side wall-clock ceiling per run (milliseconds). Upper bound for `RunOptions.configurable.runTimeoutMs`. Breach emits `cap.breached { kind: 'run-duration' }` + error `run_timeout`. Optional; hosts that advertise it MUST enforce it."
+    },
+    "maxLoopIterations": {
+      "type": "integer",
+      "minimum": 1,
+      "description": "RFC 0058. Authoritative engine-side ceiling on agent-loop iterations. Upper bound for `RunOptions.configurable.maxLoopIterations`. Breach emits `cap.breached { kind: 'loop-iterations' }` + error `loop_limit_exceeded`. Optional; hosts that advertise it MUST enforce it."
+    },
+    "maxBudgetTokens": {
+      "type": "integer",
+      "minimum": 1,
+      "description": "RFC 0084. Engine-side ceiling clamping `RunOptions.configurable.budget.maxTokens` (`min(requested, ceiling)`). Optional; only meaningful with `capabilities.budget`."
+    },
+    "maxBudgetCostUsd": {
+      "type": "number",
+      "minimum": 0,
+      "description": "RFC 0084. Engine-side cost ceiling clamping `RunOptions.configurable.budget.maxCostUsd`. Optional; only meaningful with `capabilities.budget`."
+    },
+    "maxRequestBodyBytes": {
+      "type": "integer",
+      "minimum": 1,
+      "description": "RFC 0094 §H. Maximum REST request body size (bytes) the host accepts. Optional v1 field per `capabilities.md` §3 (previously documented as reserved — the closed `limits` object made advertising it a schema-validation failure). Hosts that advertise it MUST enforce it."
+    },
+    "maxStreamAgeSec": {
+      "type": "integer",
+      "minimum": 15,
+      "description": "RFC 0247. The longest, in seconds, this host holds a live `streamRunEvents` or `streamHostEvents` stream that the consumer opened with `resumable=true`. Only for such a stream, a host advertising this facet MAY end it before the run's terminal event (a bounded close) once it has been open 15 seconds, and MUST end it by this age. It MUST end at a frame boundary, MUST flush any pending batch first, and SHOULD first send an SSE `retry:` field (milliseconds). A host MUST NOT end an opted-in stream early unless it advertises this facet, and MUST NOT end a stream opened without `resumable=true` early whatever it advertises (events.md §Resuming with `Last-Event-ID`). A bounded close is never a terminal signal; only the terminal event is. A consumer that sent `resumable=true` MUST treat any close without the terminal event (a bounded close, a network drop, an intermediary's timeout) as non-terminal, and MUST reconnect with `resumable=true` and `Last-Event-ID` set to the last `id:` it received (no header if it received none). It SHOULD wait the `retry:` delay, or its own default of about one second, and SHOULD back off when reconnects fail or close without a frame. A consumer MUST handle such a close whether or not this facet is advertised. On `hostEvents` the consumer resumes from the last durable event's `id:`; ephemeral events missed in the gap are not redelivered (events.md §Host events)."
+    }
+  },
+  "required": [
+    "clarificationRounds",
+    "schemaRounds",
+    "envelopesPerTurn"
+  ]
+}

package/spec/v2/release.json

@@ -1,6 +1,6 @@
 {
   "$comment": "RFC 0172 §D.1 — the one release identity the v2 artifacts derive from. `version` is the next corpus tag `v<version>` (the publish workflow's coordinated-release tag pattern `v*`; RFC 0172's `openwop/v2.<minor>.<patch>` spelling is amended to this at its flip). api/v2/*.yaml info.version, the suite's 2.x version and @openwop/spec-artifacts read it. Bumped by the release PR that cuts the tag, never by hand elsewhere.",
-  "version": "2.46.11",
-  "corpusTag": "v2.46.11",
-  "updated": "2026-10-15"
+  "version": "2.46.13",
+  "corpusTag": "v2.46.13",
+  "updated": "2026-10-16"
 }

package/spec/v2/surface-baseline.json

@@ -4,7 +4,7 @@
     "schemas/v2/**/*.json",
     "spec/v2/path-manifest.json"
   ],
-  "release": "2.46.11",
+  "release": "2.46.13",
   "kinds": {
     "union": [
       "channel",
@@ -18,7 +18,7 @@
       "type"
     ]
   },
-  "count": 8927,
+  "count": 8937,
   "surfaces": [
     "schemas/v2/a2a-task-state.schema.json#/contextId|type|string",
     "schemas/v2/a2a-task-state.schema.json#/interruptKind|enum-member|\"approval\"",
@@ -221,10 +221,12 @@
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]/argsMatch|enum-member|\"ignore\"",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]/argsMatch|enum-member|\"subset\"",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]/argsMatch|type|string",
+    "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]/isError|type|boolean",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]/tool|type|string",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]|closed|true",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]|property|args",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]|property|argsMatch",
+    "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]|property|isError",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]|property|response",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]|property|tool",
     "schemas/v2/agent-eval-suite.schema.json#/tasks/[]/fixtures/toolResponses/[]|required|tool",
@@ -981,6 +983,7 @@
     "schemas/v2/capabilities.schema.json#/agents/deployment|type|object",
     "schemas/v2/capabilities.schema.json#/agents/dispatchMapping|type|boolean",
     "schemas/v2/capabilities.schema.json#/agents/dispatch|type|boolean",
+    "schemas/v2/capabilities.schema.json#/agents/evalSuite/fixtureErrors|type|boolean",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite/fixtureMatching|type|boolean",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite/maxCostUsdPerSuite|type|number",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite/maxTasksPerSuite|type|integer",
@@ -999,6 +1002,7 @@
     "schemas/v2/capabilities.schema.json#/agents/evalSuite/trials|required|maxTrials",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite/trials|type|object",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite|closed|true",
+    "schemas/v2/capabilities.schema.json#/agents/evalSuite|property|fixtureErrors",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite|property|fixtureMatching",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite|property|maxCostUsdPerSuite",
     "schemas/v2/capabilities.schema.json#/agents/evalSuite|property|maxTasksPerSuite",
@@ -2321,6 +2325,7 @@
     "schemas/v2/capabilities.schema.json#/limits/maxNodeExecutions|type|integer",
     "schemas/v2/capabilities.schema.json#/limits/maxRequestBodyBytes|type|integer",
     "schemas/v2/capabilities.schema.json#/limits/maxRunDurationMs|type|integer",
+    "schemas/v2/capabilities.schema.json#/limits/maxStreamAgeSec|type|integer",
     "schemas/v2/capabilities.schema.json#/limits/schemaRounds|type|integer",
     "schemas/v2/capabilities.schema.json#/limits/since|type|string",
     "schemas/v2/capabilities.schema.json#/limits/status|enum-member|\"deprecated\"",
@@ -2341,6 +2346,7 @@
     "schemas/v2/capabilities.schema.json#/limits|property|maxNodeExecutions",
     "schemas/v2/capabilities.schema.json#/limits|property|maxRequestBodyBytes",
     "schemas/v2/capabilities.schema.json#/limits|property|maxRunDurationMs",
+    "schemas/v2/capabilities.schema.json#/limits|property|maxStreamAgeSec",
     "schemas/v2/capabilities.schema.json#/limits|property|schemaRounds",
     "schemas/v2/capabilities.schema.json#/limits|property|since",
     "schemas/v2/capabilities.schema.json#/limits|property|status",
@@ -5695,6 +5701,7 @@
     "schemas/v2/eval-summary.schema.json#/evaluatedModelClass|enum-member|\"research\"",
     "schemas/v2/eval-summary.schema.json#/evaluatedModelClass|enum-member|\"writing\"",
     "schemas/v2/eval-summary.schema.json#/evaluatedModelClass|type|string",
+    "schemas/v2/eval-summary.schema.json#/gatedAggregateScore|type|number",
     "schemas/v2/eval-summary.schema.json#/passedCount|type|integer",
     "schemas/v2/eval-summary.schema.json#/passed|type|boolean",
     "schemas/v2/eval-summary.schema.json#/pinnedNodeIds|type|array",
@@ -5734,6 +5741,7 @@
     "schemas/v2/eval-summary.schema.json#/tasks/[]/efficiency|property|toolCallCount",
     "schemas/v2/eval-summary.schema.json#/tasks/[]/efficiency|type|object",
     "schemas/v2/eval-summary.schema.json#/tasks/[]/fixtureMissCount|type|integer",
+    "schemas/v2/eval-summary.schema.json#/tasks/[]/gatedScore|type|number",
     "schemas/v2/eval-summary.schema.json#/tasks/[]/latencyMs|type|integer",
     "schemas/v2/eval-summary.schema.json#/tasks/[]/passed|type|boolean",
     "schemas/v2/eval-summary.schema.json#/tasks/[]/safetyFindings/[]/kind|type|string",
@@ -5797,6 +5805,7 @@
     "schemas/v2/eval-summary.schema.json#/tasks/[]|property|costUsd",
     "schemas/v2/eval-summary.schema.json#/tasks/[]|property|efficiency",
     "schemas/v2/eval-summary.schema.json#/tasks/[]|property|fixtureMissCount",
+    "schemas/v2/eval-summary.schema.json#/tasks/[]|property|gatedScore",
     "schemas/v2/eval-summary.schema.json#/tasks/[]|property|latencyMs",
     "schemas/v2/eval-summary.schema.json#/tasks/[]|property|passed",
     "schemas/v2/eval-summary.schema.json#/tasks/[]|property|safetyFindings",
@@ -5827,6 +5836,7 @@
     "schemas/v2/eval-summary.schema.json#|closed|true",
     "schemas/v2/eval-summary.schema.json#|property|aggregateScore",
     "schemas/v2/eval-summary.schema.json#|property|evaluatedModelClass",
+    "schemas/v2/eval-summary.schema.json#|property|gatedAggregateScore",
     "schemas/v2/eval-summary.schema.json#|property|passed",
     "schemas/v2/eval-summary.schema.json#|property|passedCount",
     "schemas/v2/eval-summary.schema.json#|property|pinnedNodeIds",