@openwop/openwop-conformance 2.46.17 → 2.46.18

package/CHANGELOG.md

@@ -1,3 +1,7 @@
+## [2.46.18] — 2026-10-18 — RFC 0245: a success return need not carry outcome
+
+- **`0245.fixture-error-replayed` no longer requires `outcome` on the success return.** `outcome` is optional on `agent.tool-returned`, and RFC 0064 §F lets a host omit it. The leg now asserts that the second return carries no `error` (and no `status: "error"` under tool hooks). This is an erratum from openwop-app's implementation (ADR 0929). No host was judged under 2.46.17.
+
 ## [2.46.17] — 2026-10-18 — RFC 0245: a recorded tool error replays as an error
 
 - **New fixture `fixtures/eval-suites/conformance-eval-fixture-error`.** One task calls `conformance.eval:lookup` twice. Its fixtures serve an `isError: true` entry, then a success.

package/dist/spec-artifacts.lock.json

@@ -1,5 +1,5 @@
 {
   "package": "@openwop/spec-artifacts",
-  "version": "2.46.17",
-  "stampSha256": "d8df0298a5a6405eca16d0a24cb7c48123d7e3e0675b337eca37ad36585e02f6"
+  "version": "2.46.18",
+  "stampSha256": "4e40e9463c8a0a9554e81e19d745f8dd90d59250be85da01cc3fc43eca1076f3"
 }

package/fixtures.md

@@ -1035,7 +1035,7 @@ Every task targets **the scripted agent `core.conformance.agents.scripted`**. It
 | `conformance-eval-fixture-miss` | `agents.evalSuite.fixtureMatching` | `missed` reports `fixtureMissCount: 1`, `passed: false` and `score: 0` (summary and every `eval.scored`); `served` passes with no miss. That no live tool ran is not observable here (RFC 0244 G2) |
 | `conformance-eval-trajectory-gate` | `agents.evalSuite.trajectory` | RFC 0246: both tasks keep `score` 1 and `aggregateScore` 1; `path-matched` carries `gatedScore` 1 and `path-wrong` `gatedScore` 0; `gatedAggregateScore` 0.5; suite `passed: false` although `aggregateScore` clears `passScore` 0.6 |
 | `conformance-eval-trajectory-gate-mean` | `agents.evalSuite.trajectory` + `agents.evalSuite.trials` | RFC 0246: two `eval.scored` with `score` 1 and `passed: false`; `trials.meanScore` 1, `gatedScore` 0, task `passed: false`; the summary carries `gatedAggregateScore` |
-| `conformance-eval-fixture-error` | `agents.evalSuite.fixtureErrors` + `agents.evalSuite.fixtureMatching` | RFC 0245: two `agent.tool-called` and two `agent.tool-returned` for `conformance.eval:lookup`, paired by `callId`; the first return carries `error` with non-empty `code` and `message` and no `outcome` (`status: "error"` under `toolHooks.prePostEvents`), the second carries `outcome`; `recovers` reports `fixtureMissCount: 0`. Without `fixtureErrors`: `400 validation_error`. That the agent saw an error and no live tool ran is not observable here (RFC 0245 G1, RFC 0244 G2) |
+| `conformance-eval-fixture-error` | `agents.evalSuite.fixtureErrors` + `agents.evalSuite.fixtureMatching` | RFC 0245: two `agent.tool-called` and two `agent.tool-returned` for `conformance.eval:lookup`, paired by `callId`; the first return carries `error` with non-empty `code` and `message` and no `outcome` (`status: "error"` under `toolHooks.prePostEvents`), the second carries no `error` (`outcome` is optional; not `status: "error"` under tool hooks); `recovers` reports `fixtureMissCount: 0`. Without `fixtureErrors`: `400 validation_error`. That the agent saw an error and no live tool ran is not observable here (RFC 0245 G1, RFC 0244 G2) |
 
 ## OAuth provider fixtures
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@openwop/openwop-conformance",
-  "version": "2.46.17",
+  "version": "2.46.18",
   "description": "Black-box conformance suite for OpenWOP hosts (v2 current; v1 scenarios retained for the frozen v1 tree).",
   "repository": {
     "type": "git",
@@ -56,6 +56,6 @@
     "@openwop/spec-artifacts": "file:../spec-artifacts"
   },
   "peerDependencies": {
-    "@openwop/spec-artifacts": "2.46.17"
+    "@openwop/spec-artifacts": "2.46.18"
   }
 }

package/schemas/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": "2.46.17",
-  "corpusTag": "v2.46.17",
+  "version": "2.46.18",
+  "corpusTag": "v2.46.18",
   "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": "08c520d1ccc23d43054ee5499276d73e18f3350d0d8825d5caa3934286e7af19",
-    "api/v2/asyncapi.yaml": "4625ddc33db126fd959e48084edca58121340042a5e1e170f87a9b4851b8b2d8",
-    "api/v2/openapi.yaml": "993bda2b91c365f8b3751fd3c827de538e39ae8e4ff3fefe2452b7b261b3c110",
+    "api/seams-v2.yaml": "7d694b3cb21105110cff218f33bdf9c931cee03c6ae54be6e3781dbc2f5789a7",
+    "api/v2/asyncapi.yaml": "5eae29f67aee9447008184f12c04673d84b7a2bed8616c5055b675d33c8a6d29",
+    "api/v2/openapi.yaml": "49bcf512e19b7441d33819c4cb2fce26de0251296d78da3599d509217041e1c1",
     "api/v2/redocly.yaml": "1e66b60e6118ad11a823bb620678be464d99dfe50a40e3e6f93ec9429b88b34c",
     "schemas/README.md": "276fc1d51eb08e14e9127b9fd27761d7046ee33c12000f3a64a801322db8bf58",
     "schemas/a2a-task-state.schema.json": "75d5049dea8bd873ff0e7546f1c60c8d36c219bec7084264360be54a8be30ae1",
@@ -208,7 +208,7 @@
     "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": "d8a41123025c45515092970fae945b4cfa238d550cdfb9549cb649e740e8b7e3",
+    "spec/v1/core-standard-manifest.json": "df4bf8218c18e6a19c7fca67cd6fee1bee705e979c687b8071f8d7e23fb5c10f",
     "spec/v1/deprecations.json": "e723bc11d7272c71789ad1303bb7f21809a8a024abf57594f9d5c35dc199ece9",
     "spec/v1/deprecations.schema.json": "3e393c405d2a41b467d8c5e3c468549078df1ce6a6d2588fc488097b95d9b55b",
     "spec/v1/end-of-support.json": "35f0bf3b70fe78b16e2a2fd81a153fc27e703f6c16376c6fe7cdf1f4f2c4e6d4",
@@ -314,9 +314,9 @@
     "spec/v2/path-manifest.json": "7b5679a4dad98c80d855361d88a98c8a4e045af15e310862a05b285b815dc9fc",
     "spec/v2/peer-dependency-aliases.json": "d10299280abee08258502925bc327293ee413e0108cd6e6ec75ff6110653308d",
     "spec/v2/profiles.json": "b382b261c7844798575f7c1b37c934f0da19846016a2337d883d4f5ef6e05479",
-    "spec/v2/release.json": "b6e1227e823697bad80029172057d535b866ca0d8faac6a1254ecfa9074b3654",
+    "spec/v2/release.json": "a8ae6035c74fcc0f2b13f1c02bee4862287840cc495840febc867cd66041c8ed",
     "spec/v2/retention-floors.json": "eaf3722d95c79947af1d4269ef85117e126518c588cfcf1a2b21b97269f51624",
-    "spec/v2/surface-baseline.json": "60061090d890cf22d91de7a4515e95814bde5690f9f9f843ee7efbdf0cdb1712"
+    "spec/v2/surface-baseline.json": "619c530713063e556abeb0f232cd905942ef8465b73936b9bf4f89ca01fe4485"
   },
-  "corpusCommit": "3d19f1af09cc0499a1bdfd35dac28707b16806a9"
+  "corpusCommit": "e4a97e99e7fb158f4091df43da0422866140af2a"
 }

package/src/lib/eval-fixture-error-witness.ts

@@ -8,8 +8,9 @@
  * an `isError: true` entry and then a success (RFC 0244 §D: first unconsumed
  * match). So the expected log is known in advance: two `agent.tool-called`
  * from the scripted agent, and two `agent.tool-returned` paired to them by
- * `callId` — the first with `error` and no `outcome`, the second with
- * `outcome`. A served error is not a miss, so `fixtureMissCount` stays 0.
+ * `callId` — the first with `error` and no `outcome`, the second with no
+ * `error` (`outcome` is optional on a success). A served error is not a miss,
+ * so `fixtureMissCount` stays 0.
  *
  * That the agent saw a tool error, and that no live tool ran, are seam-gated
  * (RFC 0245 G1, RFC 0244 G2) and not judged here.
@@ -67,8 +68,10 @@ export function judgeFixtureErrorReplayed(events: readonly Event[], summary: Rec
   if (second === undefined) {
     out.push(f(false, B, 'the second scripted call MUST have an agent.tool-returned with the same callId'));
   } else {
-    out.push(f('outcome' in second && second['outcome'] !== undefined, B, 'the call served by the following success entry MUST carry outcome'));
+    // A success is "not an error", not "carries outcome": `outcome` is optional
+    // on agent.tool-returned and RFC 0064 §F lets a host omit it (2.46.18).
     out.push(f(second['error'] === undefined, B, 'the call served by the following success entry MUST NOT carry error'));
+    if (opts.prePostEvents) out.push(f(second['status'] !== 'error', B, `under toolHooks.prePostEvents the success return MUST NOT carry status "error" (got ${JSON.stringify(second['status'])})`));
   }
 
   const tasks = Array.isArray(summary['tasks']) ? (summary['tasks'] as unknown[]).filter(isRec) : [];

package/src/scenarios/v2-eval-trajectory-trials.test.ts

@@ -50,8 +50,8 @@
  *      refused; `conformance-eval-fixture-error` validates (leg 2).
  *  11. `conformance-eval-fixture-error` (fixtureErrors + fixtureMatching) —
  *      the call served by the error entry is recorded with `error` and no
- *      `outcome` (`status: "error"` under tool hooks), the next with
- *      `outcome`, and `fixtureMissCount` is 0. The two-call sequence needs
+ *      `outcome` (`status: "error"` under tool hooks), the next with no
+ *      `error` (`outcome` optional), and `fixtureMissCount` is 0. The two-call sequence needs
  *      RFC 0244 §D's first-unconsumed selection, hence `fixtureMatching`.
  *  12. Refusal — a host advertising `agents.evalSuite` without `fixtureErrors`
  *      that lists the fixture refuses it with `400 validation_error`.