veryfront 0.1.1212 → 0.1.1214
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/esm/cli/commands/eval/command.d.ts.map +1 -1
- package/esm/cli/commands/eval/command.js +51 -0
- package/esm/deno.js +1 -1
- package/esm/src/agent/runtime/index.d.ts.map +1 -1
- package/esm/src/agent/runtime/index.js +2 -6
- package/esm/src/agent/runtime/skill-policy-enforcement.d.ts +1 -1
- package/esm/src/agent/runtime/skill-policy-enforcement.d.ts.map +1 -1
- package/esm/src/agent/runtime/skill-policy-enforcement.js +1 -8
- package/esm/src/eval/report.d.ts +2 -0
- package/esm/src/eval/report.d.ts.map +1 -1
- package/esm/src/eval/report.js +3 -1
- package/esm/src/eval/run-report.js +2 -2
- package/esm/src/html/hydration-script-builder/hydration-runtime.generated.js +1 -1
- package/esm/src/platform/adapters/index.d.ts +25 -0
- package/esm/src/platform/adapters/index.d.ts.map +1 -0
- package/esm/src/platform/adapters/index.js +23 -0
- package/esm/src/platform/adapters/security/index.d.ts +8 -0
- package/esm/src/platform/adapters/security/index.d.ts.map +1 -0
- package/esm/src/platform/adapters/security/index.js +7 -0
- package/esm/src/platform/adapters/token/factory.d.ts +10 -0
- package/esm/src/platform/adapters/token/factory.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/factory.js +79 -0
- package/esm/src/platform/adapters/token/index.d.ts +9 -0
- package/esm/src/platform/adapters/token/index.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/index.js +8 -0
- package/esm/src/platform/adapters/token/integration.d.ts +6 -0
- package/esm/src/platform/adapters/token/integration.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/integration.js +61 -0
- package/esm/src/platform/adapters/token/veryfront/adapter.d.ts +22 -0
- package/esm/src/platform/adapters/token/veryfront/adapter.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/veryfront/adapter.js +88 -0
- package/esm/src/platform/adapters/token/veryfront/api-client.d.ts +15 -0
- package/esm/src/platform/adapters/token/veryfront/api-client.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/veryfront/api-client.js +212 -0
- package/esm/src/platform/adapters/token/veryfront/index.d.ts +10 -0
- package/esm/src/platform/adapters/token/veryfront/index.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/veryfront/index.js +9 -0
- package/esm/src/platform/adapters/token/veryfront/memory-adapter.d.ts +20 -0
- package/esm/src/platform/adapters/token/veryfront/memory-adapter.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/veryfront/memory-adapter.js +52 -0
- package/esm/src/platform/adapters/token/veryfront/types.d.ts +86 -0
- package/esm/src/platform/adapters/token/veryfront/types.d.ts.map +1 -0
- package/esm/src/platform/adapters/token/veryfront/types.js +81 -0
- package/esm/src/platform/compat/http/pinned-fetch.d.ts +1 -1
- package/esm/src/rendering/ssr-globals/index.d.ts.map +1 -1
- package/esm/src/rendering/ssr-globals/index.js +18 -1
- package/esm/src/security/http/config.d.ts +5 -0
- package/esm/src/security/http/config.d.ts.map +1 -1
- package/esm/src/security/http/config.js +16 -1
- package/esm/src/security/http/derived-csp-cache.d.ts +33 -0
- package/esm/src/security/http/derived-csp-cache.d.ts.map +1 -0
- package/esm/src/security/http/derived-csp-cache.js +101 -0
- package/esm/src/security/http/derived-csp-origins.d.ts +54 -0
- package/esm/src/security/http/derived-csp-origins.d.ts.map +1 -0
- package/esm/src/security/http/derived-csp-origins.js +156 -0
- package/esm/src/security/http/response/security-handler.d.ts +1 -1
- package/esm/src/security/http/response/security-handler.d.ts.map +1 -1
- package/esm/src/security/http/response/security-handler.js +79 -4
- package/esm/src/security/sandbox/constants.d.ts +3 -0
- package/esm/src/security/sandbox/constants.d.ts.map +1 -0
- package/esm/src/security/sandbox/constants.js +2 -0
- package/esm/src/security/sandbox/deno-sandbox.d.ts +71 -0
- package/esm/src/security/sandbox/deno-sandbox.d.ts.map +1 -0
- package/esm/src/security/sandbox/deno-sandbox.js +184 -0
- package/esm/src/security/sandbox/permission-system.d.ts +11 -0
- package/esm/src/security/sandbox/permission-system.d.ts.map +1 -0
- package/esm/src/security/sandbox/permission-system.js +85 -0
- package/esm/src/server/runtime-handler/adapter-factory.d.ts +26 -0
- package/esm/src/server/runtime-handler/adapter-factory.d.ts.map +1 -1
- package/esm/src/server/runtime-handler/adapter-factory.js +19 -1
- package/esm/src/server/runtime-handler/project-runtime-context.d.ts.map +1 -1
- package/esm/src/server/runtime-handler/project-runtime-context.js +122 -1
- package/esm/src/types/server.d.ts +10 -0
- package/esm/src/types/server.d.ts.map +1 -1
- package/esm/src/utils/version-constant.d.ts +1 -1
- package/esm/src/utils/version-constant.js +1 -1
- package/esm/src/workflow/api.d.ts +2 -2
- package/package.json +8 -8
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"command.d.ts","sourceRoot":"","sources":["../../../../src/cli/commands/eval/command.ts"],"names":[],"mappings":"AAOA,OAAO,KAAK,EAAE,KAAK,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAC;AACxE,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,8BAA8B,CAAC;AACpE,OAAO,EAEL,KAAK,IAAI,EACT,KAAK,oBAAoB,EAE1B,MAAM,4BAA4B,CAAC;AACpC,OAAO,KAAK,EACV,cAAc,EACd,uBAAuB,
|
|
1
|
+
{"version":3,"file":"command.d.ts","sourceRoot":"","sources":["../../../../src/cli/commands/eval/command.ts"],"names":[],"mappings":"AAOA,OAAO,KAAK,EAAE,KAAK,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAC;AACxE,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,8BAA8B,CAAC;AACpE,OAAO,EAEL,KAAK,IAAI,EACT,KAAK,oBAAoB,EAE1B,MAAM,4BAA4B,CAAC;AACpC,OAAO,KAAK,EACV,cAAc,EACd,uBAAuB,EAIvB,0BAA0B,EAC1B,UAAU,EAGV,sBAAsB,EACtB,sBAAsB,EACtB,YAAY,EACb,MAAM,4BAA4B,CAAC;AAGpC,OAAO,EAIL,KAAK,yBAAyB,EAC/B,MAAM,uCAAuC,CAAC;AAW/C,OAAO,EACL,2BAA2B,EAG5B,MAAM,6CAA6C,CAAC;AAUrD,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,cAAc,CAAC;AAE7C,MAAM,WAAW,WAAY,SAAQ,QAAQ;IAC3C,UAAU,CAAC,EAAE,MAAM,CAAC;CACrB;AAED,UAAU,uBAAuB;IAC/B,2BAA2B,CAAC,EAAE,OAAO,2BAA2B,CAAC;CAClE;AAED,KAAK,+BAA+B,GAAG;IACrC,gBAAgB,EAAE,MAAM,CAAC;IACzB,eAAe,EAAE,MAAM,CAAC;IACxB,cAAc,EAAE,MAAM,CAAC;IACvB,kBAAkB,EAAE,MAAM,CAAC;IAC3B,iBAAiB,EAAE,MAAM,CAAC;IAC1B,oBAAoB,EAAE,MAAM,CAAC;IAC7B,oBAAoB,EAAE,MAAM,CAAC;CAC9B,CAAC;AAOF,KAAK,6BAA6B,GAAG;IACnC,aAAa,CAAC,EAAE,SAAS,MAAM,EAAE,CAAC;IAClC,KAAK,CAAC,EAAE,CAAC,EAAE,EAAE,MAAM,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC;CACvC,CAAC;AAEF,KAAK,yBAAyB,GAAG,IAAI,CAAC,0BAA0B,EAAE,eAAe,CAAC,CAAC;AAyDnF,wBAAgB,kBAAkB,CAAC,EAAE,EAAE,MAAM,GAAG,MAAM,CAErD;AAQD,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,cAAc,EAAE,EAAE,EAAE,EAAE,MAAM,GAAG,cAAc,GAAG,SAAS,CAKhG;AAED,wBAAgB,0BAA0B,CAAC,KAAK,EAAE,OAAO,GAAG,MAAM,CASjE;AA8GD,wBAAgB,oCAAoC,CAClD,MAAM,EAAE,UAAU,EAClB,YAAY,EAAE,+BAA+B,GAC5C,UAAU,CAiBZ;AAWD,wBAAsB,2BAA2B,CAC/C,cAAc,EAAE,MAAM,EACtB,OAAO,GAAE,6BAAkC,GAC1C,OAAO,CAAC,+BAA+B,GAAG,SAAS,CAAC,CAgDtD;AAED,wBAAsB,8BAA8B,CAClD,cAAc,EAAE,MAAM,EACtB,SAAS,EAAE,MAAM,OAAO,CAAC,UAAU,CAAC,GACnC,OAAO,CAAC,UAAU,CAAC,CAerB;AAED,wBAAsB,sBAAsB,CAC1C,MAAM,EAAE,UAAU,EAClB,MAAM,CAAC,EAAE,sBAAsB,GAC9B,OAAO,CAAC,UAAU,CAAC,CAGrB;AA8FD,wBAAsB,6BAA6B,CACjD,UAAU,EAAE,MAAM,EAClB,UAAU,CAAC,EAAE,MAAM,GAClB,OAAO,CAAC,yBAAyB,CAAC,CA0CpC;AAED,wBAAsB,uCAAuC,CAC3D,UAAU,EAAE,MAAM,EAClB,OAAO,EAAE,WAAW,GACnB,OAAO,CAAC,iCAAiC,GAAG,SAAS,CAAC,CAKxD;AAYD,wBAAgB,mBAAmB,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAE1D;AAED,KAAK,qBAAqB,GAAG,IAAI,CAAC,eAAe,EAAE,aAAa,GAAG,IAAI,CAAC,GAAG;IACzE,WAAW,CAAC,EAAE,MAAM,CAAC;CACtB,CAAC;AAQF,wBAAsB,sBAAsB,CAC1C,UAAU,EAAE,MAAM,EAClB,MAAM,EAAE,qBAAqB,GAAG,IAAI,GAAG,SAAS,sEAQjD;AAaD,wBAAgB,cAAc,CAAC,QAAQ,EAAE,aAAa,MA0DrD;AAED,wBAAgB,kBAAkB,CAAC,QAAQ,EAAE,aAAa,GAAG,YAAY,EAAE,CAY1E;AAyBD,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,KAAK,EAAE,OAAO,EAAE,WAAW,IACrD,qCAAqC,uBAAuB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmC3E;AAED,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,IAAI,EAAE,WAAW,GAAE,oBAAyB,IACpE,SAAS,sBAAsB;;;;;;GAiB9C;AA8ND,KAAK,yBAAyB,GAAG;IAC/B,aAAa,EAAE,MAAM,CAAC;IACtB,eAAe,EAAE,MAAM,EAAE,CAAC;IAC1B,gBAAgB,CAAC,EAAE,MAAM,CAAC;IAC1B,MAAM,EAAE,MAAM,EAAE,CAAC;CAClB,CAAC;AAEF,KAAK,iCAAiC,GAAG;IACvC,MAAM,EAAE,yBAAyB,CAAC;IAClC,MAAM,EAAE,yBAAyB,CAAC;CACnC,CAAC;AAuBF,wBAAgB,sBAAsB,CAAC,OAAO,EAAE,IAAI,CAAC,WAAW,EAAE,WAAW,CAAC,GAAG,MAAM,EAAE,CAUxF;AAED,wBAAgB,yBAAyB,CACvC,OAAO,EAAE,IAAI,CAAC,WAAW,EAAE,eAAe,CAAC,GAC1C,OAAO,CAGT;AA2BD,wBAAgB,iCAAiC,IAAI,yBAAyB,CAuB7E;AAqMD,wBAAsB,cAAc,CAClC,OAAO,EAAE,WAAW,EACpB,YAAY,GAAE,uBAA4B,GACzC,OAAO,CAAC,MAAM,GAAG,SAAS,CAAC,CA2V7B;AAED,wBAAsB,WAAW,CAAC,OAAO,EAAE,WAAW,GAAG,OAAO,CAAC,IAAI,CAAC,CAKrE"}
|
|
@@ -4,6 +4,7 @@
|
|
|
4
4
|
import * as dntShim from "../../../_dnt.shims.js";
|
|
5
5
|
import { dirname, isAbsolute, relative, resolve } from "../../../src/platform/compat/path/index.js";
|
|
6
6
|
import { isErroredToolExecutionResult, } from "../../../src/tool/index.js";
|
|
7
|
+
import { RECORD_INCOMPLETE_EXPLANATION } from "../../../src/eval/report.js";
|
|
7
8
|
import { orchestrateExtensions } from "../../../src/extensions/index.js";
|
|
8
9
|
import { createEvalReportExporterRegistry, EvalReportExporterRegistryName, } from "../../../src/extensions/eval/index.js";
|
|
9
10
|
import { createLLMProviderRegistry, LLMProviderRegistryName } from "../../../src/extensions/llm/index.js";
|
|
@@ -643,16 +644,66 @@ function printArtifact(label, path) {
|
|
|
643
644
|
function formatPercent(rate) {
|
|
644
645
|
return `${Math.round(rate * 100)}%`;
|
|
645
646
|
}
|
|
647
|
+
/** Reasons printed per metric before deferring the rest to the written report. */
|
|
648
|
+
const MAX_PRINTED_REASONS = 3;
|
|
649
|
+
/**
|
|
650
|
+
* Group gate failure explanations by the metric that produced them.
|
|
651
|
+
*
|
|
652
|
+
* A failing metric line states that something failed, and the explanation states why. The judge
|
|
653
|
+
* rubric is the clearest case: its verdict is the whole reason the eval failed, and reading it
|
|
654
|
+
* used to mean opening `summary.json`.
|
|
655
|
+
*
|
|
656
|
+
* `record.error` is dropped only when it carries the `RECORD_INCOMPLETE_EXPLANATION` stand-in and
|
|
657
|
+
* the record already reported a blocking failure. `isBlockingFailure` clears `completed` for both
|
|
658
|
+
* gate and budget severities (src/eval/runner.ts), so either one produces that stand-in, and
|
|
659
|
+
* printing it would restate the failure above it. A record error with text of its own, such as an
|
|
660
|
+
* adapter error, always prints: it is detail no other line carries.
|
|
661
|
+
*/
|
|
662
|
+
function groupGateFailureReasons(failures) {
|
|
663
|
+
const derived = new Set(failures.filter((failure) => failure.name !== "record.error").map((failure) => failure.recordId));
|
|
664
|
+
const byMetric = new Map();
|
|
665
|
+
const unattached = [];
|
|
666
|
+
for (const failure of failures) {
|
|
667
|
+
if (!failure.explanation)
|
|
668
|
+
continue;
|
|
669
|
+
if (failure.name === "record.error" &&
|
|
670
|
+
failure.explanation === RECORD_INCOMPLETE_EXPLANATION &&
|
|
671
|
+
derived.has(failure.recordId))
|
|
672
|
+
continue;
|
|
673
|
+
const reason = `${failure.exampleId}: ${failure.explanation}`;
|
|
674
|
+
if (failure.name === "record.error") {
|
|
675
|
+
unattached.push(reason);
|
|
676
|
+
continue;
|
|
677
|
+
}
|
|
678
|
+
const existing = byMetric.get(failure.name);
|
|
679
|
+
if (existing)
|
|
680
|
+
existing.push(reason);
|
|
681
|
+
else
|
|
682
|
+
byMetric.set(failure.name, [reason]);
|
|
683
|
+
}
|
|
684
|
+
return { byMetric, unattached };
|
|
685
|
+
}
|
|
686
|
+
function printFailureReasons(reasons) {
|
|
687
|
+
for (const reason of reasons.slice(0, MAX_PRINTED_REASONS)) {
|
|
688
|
+
printLine(` ${dim(reason)}`);
|
|
689
|
+
}
|
|
690
|
+
const hidden = reasons.length - MAX_PRINTED_REASONS;
|
|
691
|
+
if (hidden > 0)
|
|
692
|
+
printLine(` ${dim(`and ${hidden} more, see the report`)}`);
|
|
693
|
+
}
|
|
646
694
|
function printReport(report, options = {}) {
|
|
647
695
|
printHeader("Eval", options.name ?? report.definitionId);
|
|
648
696
|
printHeader("Target", report.target);
|
|
649
697
|
printHeader("Result", `${report.summary.passed}/${report.summary.records} passed (${formatPercent(report.summary.passRate)})`);
|
|
698
|
+
const reasons = groupGateFailureReasons(report.summary.gateFailures ?? []);
|
|
650
699
|
if (report.summary.metrics.length > 0)
|
|
651
700
|
printBlankLine();
|
|
652
701
|
for (const metric of report.summary.metrics) {
|
|
653
702
|
const total = metric.passed + metric.failed;
|
|
654
703
|
printLine(`${brand("●")} ${metric.label ?? metric.name}: ${metric.passed}/${total} passed (${formatPercent(metric.passRate)})`);
|
|
704
|
+
printFailureReasons(reasons.byMetric.get(metric.name) ?? []);
|
|
655
705
|
}
|
|
706
|
+
printFailureReasons(reasons.unattached);
|
|
656
707
|
const { baseline } = options;
|
|
657
708
|
const notes = (report.exports ?? []).length > 0 || baseline !== undefined;
|
|
658
709
|
if (notes)
|
package/esm/deno.js
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/src/agent/runtime/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EACL,KAAK,WAAW,EAEhB,KAAK,6BAA6B,EAClC,KAAK,aAAa,EAGlB,KAAK,OAAO,EAIZ,KAAK,QAAQ,EAGd,MAAM,aAAa,CAAC;AAIrB,OAAO,EAAqB,KAAK,MAAM,EAAE,MAAM,oBAAoB,CAAC;AA4EpE,OAAO,EAAE,cAAc,EAAE,iBAAiB,EAAE,OAAO,EAAE,MAAM,gBAAgB,CAAC;AAC5E,OAAO,EACL,qBAAqB,EACrB,iBAAiB,EACjB,iBAAiB,EACjB,uBAAuB,EACvB,iBAAiB,EACjB,mBAAmB,GACpB,MAAM,qBAAqB,CAAC;AAC7B,YAAY,EACV,8BAA8B,EAC9B,gBAAgB,EAChB,wBAAwB,GACzB,MAAM,qBAAqB,CAAC;AAC7B,OAAO,EACL,qBAAqB,EACrB,iBAAiB,EACjB,aAAa,EACb,aAAa,EACb,qBAAqB,GACtB,MAAM,mBAAmB,CAAC;AAC3B,YAAY,EAAE,cAAc,EAAE,eAAe,EAAE,MAAM,mBAAmB,CAAC;AACzE,OAAO,EACL,sBAAsB,EACtB,KAAK,yBAAyB,EAC9B,KAAK,0BAA0B,EAC/B,KAAK,mBAAmB,EACxB,0BAA0B,EAC1B,iCAAiC,EACjC,6BAA6B,GAC9B,MAAM,2BAA2B,CAAC;AACnC,OAAO,EAAE,eAAe,EAAE,WAAW,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC;AAChF,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;AAG5E,YAAY,EACV,mBAAmB,EACnB,eAAe,EACf,iBAAiB,GAClB,MAAM,0BAA0B,CAAC;AAClC,OAAO,EACL,iBAAiB,EACjB,kBAAkB,EAClB,mBAAmB,EACnB,sBAAsB,GACvB,MAAM,gBAAgB,CAAC;AACxB,OAAO,EACL,4BAA4B,EAC5B,6BAA6B,EAC7B,2BAA2B,EAC3B,2BAA2B,EAC3B,gCAAgC,EAChC,4BAA4B,EAC5B,2BAA2B,EAC3B,6BAA6B,EAC7B,KAAK,+BAA+B,GACrC,MAAM,+BAA+B,CAAC;AAuBvC,OAAO,EACL,kBAAkB,EAClB,kBAAkB,EAClB,KAAK,iBAAiB,GACvB,MAAM,+BAA+B,CAAC;AAwevC,+BAA+B;AAC/B,qBAAa,YAAY;IACvB,OAAO,CAAC,EAAE,CAAS;IACnB,OAAO,CAAC,MAAM,CAAc;IAC5B,OAAO,CAAC,MAAM,CAAkB;IAChC,OAAO,CAAC,MAAM,CAAuB;gBAEzB,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,WAAW;IAW3C;;;;;;;OAOG;YACW,mBAAmB;YAMnB,qBAAqB;YAcrB,mBAAmB;YAsBnB,gBAAgB;IAS9B,OAAO,CAAC,8BAA8B;IA0BtC;;OAEG;IACG,QAAQ,CACZ,KAAK,EAAE,MAAM,GAAG,OAAO,EAAE,EACzB,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACjC,aAAa,CAAC,EAAE,MAAM,EACtB,uBAAuB,CAAC,EAAE,MAAM,EAChC,WAAW,CAAC,EAAE,WAAW,EACzB,OAAO,CAAC,EAAE;QACR,gBAAgB,CAAC,EAAE,6BAA6B,CAAC;QACjD,sBAAsB,CAAC,EAAE,OAAO,CAAC;KAClC,GACA,OAAO,CAAC,aAAa,CAAC;IAyDzB;;;OAGG;IACG,MAAM,CACV,QAAQ,EAAE,OAAO,EAAE,EACnB,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACjC,SAAS,CAAC,EAAE;QACV,UAAU,CAAC,EAAE,CAAC,QAAQ,EAAE,QAAQ,KAAK,IAAI,CAAC;QAC1C,OAAO,CAAC,EAAE,CAAC,KAAK,EAAE,MAAM,KAAK,IAAI,CAAC;QAClC,QAAQ,CAAC,EAAE,CAAC,QAAQ,EAAE,aAAa,KAAK,IAAI,CAAC;KAC9C,EACD,aAAa,CAAC,EAAE,MAAM,EACtB,uBAAuB,CAAC,EAAE,MAAM,EAChC,WAAW,CAAC,EAAE,WAAW,GACxB,OAAO,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC;IAyJtC;;OAEG;YACW,gBAAgB;
|
|
1
|
+
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/src/agent/runtime/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EACL,KAAK,WAAW,EAEhB,KAAK,6BAA6B,EAClC,KAAK,aAAa,EAGlB,KAAK,OAAO,EAIZ,KAAK,QAAQ,EAGd,MAAM,aAAa,CAAC;AAIrB,OAAO,EAAqB,KAAK,MAAM,EAAE,MAAM,oBAAoB,CAAC;AA4EpE,OAAO,EAAE,cAAc,EAAE,iBAAiB,EAAE,OAAO,EAAE,MAAM,gBAAgB,CAAC;AAC5E,OAAO,EACL,qBAAqB,EACrB,iBAAiB,EACjB,iBAAiB,EACjB,uBAAuB,EACvB,iBAAiB,EACjB,mBAAmB,GACpB,MAAM,qBAAqB,CAAC;AAC7B,YAAY,EACV,8BAA8B,EAC9B,gBAAgB,EAChB,wBAAwB,GACzB,MAAM,qBAAqB,CAAC;AAC7B,OAAO,EACL,qBAAqB,EACrB,iBAAiB,EACjB,aAAa,EACb,aAAa,EACb,qBAAqB,GACtB,MAAM,mBAAmB,CAAC;AAC3B,YAAY,EAAE,cAAc,EAAE,eAAe,EAAE,MAAM,mBAAmB,CAAC;AACzE,OAAO,EACL,sBAAsB,EACtB,KAAK,yBAAyB,EAC9B,KAAK,0BAA0B,EAC/B,KAAK,mBAAmB,EACxB,0BAA0B,EAC1B,iCAAiC,EACjC,6BAA6B,GAC9B,MAAM,2BAA2B,CAAC;AACnC,OAAO,EAAE,eAAe,EAAE,WAAW,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC;AAChF,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;AAG5E,YAAY,EACV,mBAAmB,EACnB,eAAe,EACf,iBAAiB,GAClB,MAAM,0BAA0B,CAAC;AAClC,OAAO,EACL,iBAAiB,EACjB,kBAAkB,EAClB,mBAAmB,EACnB,sBAAsB,GACvB,MAAM,gBAAgB,CAAC;AACxB,OAAO,EACL,4BAA4B,EAC5B,6BAA6B,EAC7B,2BAA2B,EAC3B,2BAA2B,EAC3B,gCAAgC,EAChC,4BAA4B,EAC5B,2BAA2B,EAC3B,6BAA6B,EAC7B,KAAK,+BAA+B,GACrC,MAAM,+BAA+B,CAAC;AAuBvC,OAAO,EACL,kBAAkB,EAClB,kBAAkB,EAClB,KAAK,iBAAiB,GACvB,MAAM,+BAA+B,CAAC;AAwevC,+BAA+B;AAC/B,qBAAa,YAAY;IACvB,OAAO,CAAC,EAAE,CAAS;IACnB,OAAO,CAAC,MAAM,CAAc;IAC5B,OAAO,CAAC,MAAM,CAAkB;IAChC,OAAO,CAAC,MAAM,CAAuB;gBAEzB,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,WAAW;IAW3C;;;;;;;OAOG;YACW,mBAAmB;YAMnB,qBAAqB;YAcrB,mBAAmB;YAsBnB,gBAAgB;IAS9B,OAAO,CAAC,8BAA8B;IA0BtC;;OAEG;IACG,QAAQ,CACZ,KAAK,EAAE,MAAM,GAAG,OAAO,EAAE,EACzB,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACjC,aAAa,CAAC,EAAE,MAAM,EACtB,uBAAuB,CAAC,EAAE,MAAM,EAChC,WAAW,CAAC,EAAE,WAAW,EACzB,OAAO,CAAC,EAAE;QACR,gBAAgB,CAAC,EAAE,6BAA6B,CAAC;QACjD,sBAAsB,CAAC,EAAE,OAAO,CAAC;KAClC,GACA,OAAO,CAAC,aAAa,CAAC;IAyDzB;;;OAGG;IACG,MAAM,CACV,QAAQ,EAAE,OAAO,EAAE,EACnB,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACjC,SAAS,CAAC,EAAE;QACV,UAAU,CAAC,EAAE,CAAC,QAAQ,EAAE,QAAQ,KAAK,IAAI,CAAC;QAC1C,OAAO,CAAC,EAAE,CAAC,KAAK,EAAE,MAAM,KAAK,IAAI,CAAC;QAClC,QAAQ,CAAC,EAAE,CAAC,QAAQ,EAAE,aAAa,KAAK,IAAI,CAAC;KAC9C,EACD,aAAa,CAAC,EAAE,MAAM,EACtB,uBAAuB,CAAC,EAAE,MAAM,EAChC,WAAW,CAAC,EAAE,WAAW,GACxB,OAAO,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC;IAyJtC;;OAEG;YACW,gBAAgB;IAimB9B;;;;OAIG;YACW,yBAAyB;IAsmBvC;;OAEG;YACW,eAAe;IAgC7B;;OAEG;YACW,mBAAmB;IAOjC;;OAEG;IACH,OAAO,CAAC,eAAe;IAKvB,OAAO,CAAC,kBAAkB;IAY1B,OAAO,CAAC,sBAAsB;IAoB9B;;OAEG;IACH,SAAS,IAAI,MAAM,CAAC,OAAO,CAAC;IAI5B;;OAEG;IACG,cAAc,IAAI,OAAO,CAAC;QAC9B,aAAa,EAAE,MAAM,CAAC;QACtB,eAAe,EAAE,MAAM,CAAC;QACxB,IAAI,EAAE,MAAM,CAAC;KACd,CAAC;IAIF;;OAEG;IACG,WAAW,IAAI,OAAO,CAAC,IAAI,CAAC;CAGnC"}
|
|
@@ -876,8 +876,6 @@ export class AgentRuntime {
|
|
|
876
876
|
}
|
|
877
877
|
this.status = "tool_execution";
|
|
878
878
|
addSpanEvent(loopSpan, "tool_execution_start", { count: response.toolCalls.length });
|
|
879
|
-
const mustLoadSkillFirstForStep = !skillState.activeSkillPolicy &&
|
|
880
|
-
response.toolCalls.some((tc) => tc.toolName === LOAD_SKILL_TOOL_ID);
|
|
881
879
|
for (const tc of response.toolCalls) {
|
|
882
880
|
throwIfAborted(abortSignal);
|
|
883
881
|
const toolCall = {
|
|
@@ -1011,7 +1009,7 @@ export class AgentRuntime {
|
|
|
1011
1009
|
toolCalls.push(toolCall);
|
|
1012
1010
|
return;
|
|
1013
1011
|
}
|
|
1014
|
-
const policyCheck = enforceSkillPolicy(tc.toolName, skillState.activeSkillPolicy,
|
|
1012
|
+
const policyCheck = enforceSkillPolicy(tc.toolName, skillState.activeSkillPolicy, {
|
|
1015
1013
|
activeSkillId: skillState.activeSkillId,
|
|
1016
1014
|
hasSubmittedFormInput: skillState.hasSubmittedFormInput,
|
|
1017
1015
|
skillToolAvailability: skillState.activeSkillToolAvailability,
|
|
@@ -1350,8 +1348,6 @@ export class AgentRuntime {
|
|
|
1350
1348
|
}
|
|
1351
1349
|
this.status = "tool_execution";
|
|
1352
1350
|
const streamedToolCalls = Array.from(state.toolCalls.values());
|
|
1353
|
-
const mustLoadSkillFirstForStep = !skillState.activeSkillPolicy &&
|
|
1354
|
-
streamedToolCalls.some((tc) => tc.name === LOAD_SKILL_TOOL_ID);
|
|
1355
1351
|
for (const tc of streamedToolCalls) {
|
|
1356
1352
|
throwIfAborted(abortSignal);
|
|
1357
1353
|
if (isRecoverablePlaceholderToolCall(tc)) {
|
|
@@ -1512,7 +1508,7 @@ export class AgentRuntime {
|
|
|
1512
1508
|
await this.recordToolError(toolCall, toolNotVisibleError(tc.name), controller, encoder, currentMessages, toolCalls);
|
|
1513
1509
|
continue;
|
|
1514
1510
|
}
|
|
1515
|
-
const policyCheck = enforceSkillPolicy(tc.name, skillState.activeSkillPolicy,
|
|
1511
|
+
const policyCheck = enforceSkillPolicy(tc.name, skillState.activeSkillPolicy, {
|
|
1516
1512
|
activeSkillId: skillState.activeSkillId,
|
|
1517
1513
|
hasSubmittedFormInput: skillState.hasSubmittedFormInput,
|
|
1518
1514
|
skillToolAvailability: skillState.activeSkillToolAvailability,
|
|
@@ -41,5 +41,5 @@ export type SkillPolicyOptions = {
|
|
|
41
41
|
};
|
|
42
42
|
/** Identify a valid skill-body activation call without confusing reference reads for activation. */
|
|
43
43
|
export declare function isSkillBodyLoadRequest(toolName: string, input: unknown): boolean;
|
|
44
|
-
export declare function enforceSkillPolicy(toolName: string, activeSkillPolicy: string[] | undefined,
|
|
44
|
+
export declare function enforceSkillPolicy(toolName: string, activeSkillPolicy: string[] | undefined, options?: SkillPolicyOptions): SkillPolicyResult;
|
|
45
45
|
//# sourceMappingURL=skill-policy-enforcement.d.ts.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"skill-policy-enforcement.d.ts","sourceRoot":"","sources":["../../../../src/src/agent/runtime/skill-policy-enforcement.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,aAAa,CAAC;AAC3C,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAE1D,OAAO,EAEL,KAAK,qBAAqB,EAE3B,MAAM,8BAA8B,CAAC;AAiBtC,OAAO,EAEL,KAAK,wBAAwB,EAC9B,MAAM,iCAAiC,CAAC;AAKzC,eAAO,MAAM,kBAAkB,eAAe,CAAC;AAC/C,eAAO,MAAM,kBAAkB,eAAe,CAAC;AAC/C,eAAO,MAAM,oBAAoB,iBAAiB,CAAC;AACnD,eAAO,MAAM,gCAAgC,gCAAgC,CAAC;AAI9E,eAAO,MAAM,gCAAgC,EAAE,qBAI7C,CAAC;AAgFH,MAAM,MAAM,gBAAgB,GAAG;IAC7B,aAAa,EAAE,MAAM,GAAG,SAAS,CAAC;IAClC,iBAAiB,EAAE,MAAM,EAAE,GAAG,SAAS,CAAC;IACxC,2BAA2B,EAAE,qBAAqB,CAAC;IACnD,8BAA8B,EAAE,wBAAwB,GAAG,SAAS,CAAC;CACtE,CAAC;
|
|
1
|
+
{"version":3,"file":"skill-policy-enforcement.d.ts","sourceRoot":"","sources":["../../../../src/src/agent/runtime/skill-policy-enforcement.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,aAAa,CAAC;AAC3C,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAE1D,OAAO,EAEL,KAAK,qBAAqB,EAE3B,MAAM,8BAA8B,CAAC;AAiBtC,OAAO,EAEL,KAAK,wBAAwB,EAC9B,MAAM,iCAAiC,CAAC;AAKzC,eAAO,MAAM,kBAAkB,eAAe,CAAC;AAC/C,eAAO,MAAM,kBAAkB,eAAe,CAAC;AAC/C,eAAO,MAAM,oBAAoB,iBAAiB,CAAC;AACnD,eAAO,MAAM,gCAAgC,gCAAgC,CAAC;AAI9E,eAAO,MAAM,gCAAgC,EAAE,qBAI7C,CAAC;AAgFH,MAAM,MAAM,gBAAgB,GAAG;IAC7B,aAAa,EAAE,MAAM,GAAG,SAAS,CAAC;IAClC,iBAAiB,EAAE,MAAM,EAAE,GAAG,SAAS,CAAC;IACxC,2BAA2B,EAAE,qBAAqB,CAAC;IACnD,8BAA8B,EAAE,wBAAwB,GAAG,SAAS,CAAC;CACtE,CAAC;AAEF,wBAAgB,mCAAmC,CACjD,QAAQ,EAAE,SAAS,OAAO,EAAE,GAC3B,gBAAgB,CAgBlB;AAED,wBAAgB,cAAc,CAAC,MAAM,EAAE,OAAO,GAAG,MAAM,GAAG,SAAS,CAIlE;AAED,wBAAgB,kBAAkB,CAAC,MAAM,EAAE,OAAO,GAAG,MAAM,EAAE,GAAG,SAAS,CAiCxE;AAiCD,wBAAgB,4BAA4B,CAC1C,MAAM,EAAE,OAAO,GACd,qBAAqB,GAAG,SAAS,CAkBnC;AAED,yFAAyF;AACzF,wBAAgB,0BAA0B,CACxC,OAAO,EAAE,gBAAgB,EACzB,MAAM,EAAE,OAAO,GACd,gBAAgB,CAiBlB;AAUD,wBAAgB,0BAA0B,CAAC,MAAM,EAAE,OAAO,GAAG,OAAO,CAgBnE;AAYD,wBAAgB,2BAA2B,CAAC,QAAQ,EAAE,SAAS,OAAO,EAAE,GAAG,OAAO,CAUjF;AAED,wBAAgB,kCAAkC,CAChD,KAAK,EAAE,SAAS,cAAc,EAAE,EAChC,QAAQ,EAAE,SAAS,OAAO,EAAE,EAC5B,cAAc,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EACxC,WAAW,CAAC,EAAE;IACZ,EAAE,CAAC,EAAE,MAAM,CAAC;IACZ,gBAAgB,CAAC,EAAE,qBAAqB,CAAC;CAC1C,GACA,cAAc,EAAE,CAkClB;AAED,wBAAgB,8BAA8B,CAC5C,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,OAAO,EACf,iBAAiB,EAAE,MAAM,EAAE,GAAG,SAAS,GACtC,MAAM,EAAE,GAAG,SAAS,CAStB;AAED,wBAAgB,8BAA8B,CAC5C,iBAAiB,EAAE,MAAM,EAAE,GAAG,SAAS,GACtC,MAAM,EAAE,GAAG,SAAS,CAMtB;AAED,MAAM,MAAM,iBAAiB,GACzB;IAAE,OAAO,EAAE,IAAI,CAAA;CAAE,GACjB;IAAE,OAAO,EAAE,KAAK,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC;AAEtC,MAAM,MAAM,kBAAkB,GAAG;IAC/B,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,qBAAqB,CAAC,EAAE,qBAAqB,CAAC;IAC9C,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,SAAS,CAAC,EAAE,OAAO,CAAC;CACrB,CAAC;AA2BF,oGAAoG;AACpG,wBAAgB,sBAAsB,CAAC,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,OAAO,GAAG,OAAO,CAKhF;AAED,wBAAgB,kBAAkB,CAChC,QAAQ,EAAE,MAAM,EAChB,iBAAiB,EAAE,MAAM,EAAE,GAAG,SAAS,EACvC,OAAO,GAAE,kBAAuB,GAC/B,iBAAiB,CAoCnB"}
|
|
@@ -86,10 +86,6 @@ function isSkillActivationResult(result) {
|
|
|
86
86
|
instructions.length <= SKILL_DOCUMENT_MAX_CHARACTERS &&
|
|
87
87
|
isWellFormedUtf16(instructions);
|
|
88
88
|
}
|
|
89
|
-
function getSkillActivationRequiredError(toolName) {
|
|
90
|
-
return `Tool "${toolName}" cannot run before load_skill succeeds in the same step. ` +
|
|
91
|
-
`Call "${LOAD_SKILL_TOOL_ID}" first to establish the active skill context.`;
|
|
92
|
-
}
|
|
93
89
|
export function hydrateActiveSkillStateFromMessages(messages) {
|
|
94
90
|
let state = {
|
|
95
91
|
activeSkillId: undefined,
|
|
@@ -300,10 +296,7 @@ export function isSkillBodyLoadRequest(toolName, input) {
|
|
|
300
296
|
const file = readToolResultOwnDataProperty(input, "file");
|
|
301
297
|
return typeof skillId === "string" && skillId.length > 0 && file === undefined;
|
|
302
298
|
}
|
|
303
|
-
export function enforceSkillPolicy(toolName, activeSkillPolicy,
|
|
304
|
-
if (mustLoadSkillFirst && toolName !== LOAD_SKILL_TOOL_ID) {
|
|
305
|
-
return { allowed: false, error: getSkillActivationRequiredError(toolName) };
|
|
306
|
-
}
|
|
299
|
+
export function enforceSkillPolicy(toolName, activeSkillPolicy, options = {}) {
|
|
307
300
|
if (options.hasSubmittedFormInput === true &&
|
|
308
301
|
POST_SUBMITTED_FORM_INPUT_BLOCKED_TOOL_IDS.has(toolName)) {
|
|
309
302
|
return {
|
package/esm/src/eval/report.d.ts
CHANGED
|
@@ -1,6 +1,8 @@
|
|
|
1
1
|
import type { EvalDataset, EvalDefinition, EvalExample, EvalRecord, EvalReport, EvalReportDatasetMetadata, EvalReportMetadata, EvalReportSummary } from "./types.js";
|
|
2
2
|
/** Additive eval report contract version written by new reports and summary artifacts. */
|
|
3
3
|
export declare const EVAL_REPORT_SCHEMA_VERSION = 2;
|
|
4
|
+
/** Stand-in explanation when a record is incomplete but carries no error of its own. */
|
|
5
|
+
export declare const RECORD_INCOMPLETE_EXPLANATION = "Record did not complete.";
|
|
4
6
|
/** Create stable dataset metadata for report consumers and CI artifacts. */
|
|
5
7
|
export declare function createEvalDatasetMetadata(dataset: EvalDataset, examples: EvalExample[]): Promise<EvalReportDatasetMetadata>;
|
|
6
8
|
/** Summarize eval records into pass/fail and metric aggregates. */
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"report.d.ts","sourceRoot":"","sources":["../../../src/src/eval/report.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACV,WAAW,EACX,cAAc,EAEd,WAAW,EAMX,UAAU,EACV,UAAU,EACV,yBAAyB,EACzB,kBAAkB,EAClB,iBAAiB,EAElB,MAAM,YAAY,CAAC;AAGpB,0FAA0F;AAC1F,eAAO,MAAM,0BAA0B,IAAI,CAAC;
|
|
1
|
+
{"version":3,"file":"report.d.ts","sourceRoot":"","sources":["../../../src/src/eval/report.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACV,WAAW,EACX,cAAc,EAEd,WAAW,EAMX,UAAU,EACV,UAAU,EACV,yBAAyB,EACzB,kBAAkB,EAClB,iBAAiB,EAElB,MAAM,YAAY,CAAC;AAGpB,0FAA0F;AAC1F,eAAO,MAAM,0BAA0B,IAAI,CAAC;AAE5C,wFAAwF;AACxF,eAAO,MAAM,6BAA6B,6BAA6B,CAAC;AAiDxE,4EAA4E;AAC5E,wBAAsB,yBAAyB,CAC7C,OAAO,EAAE,WAAW,EACpB,QAAQ,EAAE,WAAW,EAAE,GACtB,OAAO,CAAC,yBAAyB,CAAC,CAQpC;AAmOD,mEAAmE;AACnE,wBAAgB,oBAAoB,CAAC,OAAO,EAAE,UAAU,EAAE,GAAG,iBAAiB,CAoB7E;AAED,oEAAoE;AACpE,wBAAgB,gBAAgB,CAAC,KAAK,EAAE;IACtC,UAAU,EAAE,cAAc,CAAC;IAC3B,OAAO,EAAE,UAAU,EAAE,CAAC;IACtB,KAAK,EAAE,MAAM,CAAC;IACd,SAAS,EAAE,IAAI,CAAC;IAChB,OAAO,EAAE,IAAI,CAAC;IACd,OAAO,CAAC,EAAE,yBAAyB,CAAC;IACpC,QAAQ,CAAC,EAAE,kBAAkB,CAAC;CAC/B,GAAG,UAAU,CAeb"}
|
package/esm/src/eval/report.js
CHANGED
|
@@ -1,6 +1,8 @@
|
|
|
1
1
|
import { computeHash } from "../utils/index.js";
|
|
2
2
|
/** Additive eval report contract version written by new reports and summary artifacts. */
|
|
3
3
|
export const EVAL_REPORT_SCHEMA_VERSION = 2;
|
|
4
|
+
/** Stand-in explanation when a record is incomplete but carries no error of its own. */
|
|
5
|
+
export const RECORD_INCOMPLETE_EXPLANATION = "Record did not complete.";
|
|
4
6
|
const USAGE_NUMERIC_KEYS = [
|
|
5
7
|
"inputTokens",
|
|
6
8
|
"outputTokens",
|
|
@@ -149,7 +151,7 @@ function createRecordFailure(record) {
|
|
|
149
151
|
name: "record.error",
|
|
150
152
|
family: "check",
|
|
151
153
|
severity: "gate",
|
|
152
|
-
explanation: record.error ??
|
|
154
|
+
explanation: record.error ?? RECORD_INCOMPLETE_EXPLANATION,
|
|
153
155
|
};
|
|
154
156
|
}
|
|
155
157
|
function createResultFailure(record, result) {
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import { join, relative } from "../platform/compat/path/index.js";
|
|
2
2
|
import { compareEvalReports } from "./baseline.js";
|
|
3
3
|
import { compareEvalModelReports, createEvalModelComparisonMarkdown } from "./model-comparison.js";
|
|
4
|
-
import { EVAL_REPORT_SCHEMA_VERSION } from "./report.js";
|
|
4
|
+
import { EVAL_REPORT_SCHEMA_VERSION, RECORD_INCOMPLETE_EXPLANATION } from "./report.js";
|
|
5
5
|
import { createEvalRunId } from "./run-id.js";
|
|
6
6
|
import { INVALID_ARGUMENT } from "../errors/index.js";
|
|
7
7
|
function createEvalReportDirTimestamp(runId) {
|
|
@@ -245,7 +245,7 @@ function createJunitXml(report) {
|
|
|
245
245
|
for (const record of report.records) {
|
|
246
246
|
const failures = blockingResults(record);
|
|
247
247
|
const recordFailure = !record.completed || record.error
|
|
248
|
-
? record.error ??
|
|
248
|
+
? record.error ?? RECORD_INCOMPLETE_EXPLANATION
|
|
249
249
|
: undefined;
|
|
250
250
|
const attrs = `classname="${xmlEscape(report.definitionId)}" name="${xmlEscape(testcaseName(record))}" time="${(record.durationMs / 1000).toFixed(3)}"`;
|
|
251
251
|
if (failures.length === 0 && recordFailure === undefined) {
|