@tangle-network/agent-eval 0.200.1 → 0.202.0
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/CHANGELOG.md +9 -0
- package/dist/adapters/http.d.ts +8 -9
- package/dist/adapters/http.d.ts.map +1 -1
- package/dist/adapters/http.js +2 -2
- package/dist/adapters/http.js.map +1 -1
- package/dist/agent-profile-DDfIjdyS.d.ts.map +1 -1
- package/dist/analyst/index.d.ts +76 -76
- package/dist/analyst/index.d.ts.map +1 -1
- package/dist/analyst/index.js +6 -6
- package/dist/{analyze-runs-DFbp2gB8.js → analyze-runs-BrIibNd0.js} +47 -42
- package/dist/{analyze-runs-DFbp2gB8.js.map → analyze-runs-BrIibNd0.js.map} +1 -1
- package/dist/{ast-CP9ae9B0.js → ast-DJ6Bc7ab.js} +5 -4
- package/dist/{ast-CP9ae9B0.js.map → ast-DJ6Bc7ab.js.map} +1 -1
- package/dist/authenticity/index.d.ts +14 -15
- package/dist/authenticity/index.d.ts.map +1 -1
- package/dist/{baseline-BC-eBZ7U.js → baseline-CVKsfWTg.js} +2 -2
- package/dist/{baseline-BC-eBZ7U.js.map → baseline-CVKsfWTg.js.map} +1 -1
- package/dist/{benchmark-BeQJyrRC.js → benchmark-BywPeAUo.js} +4 -3
- package/dist/{benchmark-BeQJyrRC.js.map → benchmark-BywPeAUo.js.map} +1 -1
- package/dist/{benchmark-command-EBmuMUBo.js → benchmark-command-DFs_i62N.js} +24 -23
- package/dist/{benchmark-command-EBmuMUBo.js.map → benchmark-command-DFs_i62N.js.map} +1 -1
- package/dist/benchmarks/index.d.ts +13 -13
- package/dist/benchmarks/index.d.ts.map +1 -1
- package/dist/benchmarks/index.js +3 -3
- package/dist/{bounded-process-BBZob7vl.js → bounded-process-Cj9YjXbE.js} +3 -3
- package/dist/{bounded-process-BBZob7vl.js.map → bounded-process-Cj9YjXbE.js.map} +1 -1
- package/dist/builder-eval/index.d.ts +16 -17
- package/dist/builder-eval/index.d.ts.map +1 -1
- package/dist/builder-eval/index.js +12 -10
- package/dist/builder-eval/index.js.map +1 -1
- package/dist/campaign/index.d.ts +2 -2
- package/dist/campaign/index.js +12 -12
- package/dist/{campaign-L391bmir.js → campaign-XXmOU0R0.js} +265 -27
- package/dist/campaign-XXmOU0R0.js.map +1 -0
- package/dist/{campaign-evidence-jL_U7Ikv.js → campaign-evidence-Dt_rWPk2.js} +4 -4
- package/dist/{campaign-evidence-jL_U7Ikv.js.map → campaign-evidence-Dt_rWPk2.js.map} +1 -1
- package/dist/{chat-client-CJORSG_J.js → chat-client-DUhWNQQS.js} +5 -4
- package/dist/{chat-client-CJORSG_J.js.map → chat-client-DUhWNQQS.js.map} +1 -1
- package/dist/{chat-json-call-DnlDjpco.js → chat-json-call-Dx_sHzqm.js} +2 -2
- package/dist/{chat-json-call-DnlDjpco.js.map → chat-json-call-Dx_sHzqm.js.map} +1 -1
- package/dist/{claim-V9Zqbq0J.js → claim-BXcZw-jo.js} +2 -2
- package/dist/{claim-V9Zqbq0J.js.map → claim-BXcZw-jo.js.map} +1 -1
- package/dist/cli.js +12 -11
- package/dist/cli.js.map +1 -1
- package/dist/contract/index.d.ts +3 -3
- package/dist/contract/index.js +22 -19
- package/dist/contract/index.js.map +1 -1
- package/dist/{default-registry-xMLX160E.js → default-registry-D4qXx2aG.js} +2 -2
- package/dist/{default-registry-xMLX160E.js.map → default-registry-D4qXx2aG.js.map} +1 -1
- package/dist/{define-agent-eval-7q_UwsTt.js → define-agent-eval-CJN9dTQw.js} +10 -9
- package/dist/{define-agent-eval-7q_UwsTt.js.map → define-agent-eval-CJN9dTQw.js.map} +1 -1
- package/dist/{define-agent-eval-1YuvJpOs.d.ts → define-agent-eval-DQXLPqWc.d.ts} +2 -2
- package/dist/define-agent-eval-DQXLPqWc.d.ts.map +1 -0
- package/dist/{descriptive-1V17A-qa.js → descriptive-D9LUFk1-.js} +2 -2
- package/dist/{descriptive-1V17A-qa.js.map → descriptive-D9LUFk1-.js.map} +1 -1
- package/dist/diagnosis/index.d.ts +7 -7
- package/dist/diagnosis/index.d.ts.map +1 -1
- package/dist/diagnosis/index.js +2 -2
- package/dist/diagnosis/index.js.map +1 -1
- package/dist/{dspy-rlm-engine-B5T43BkZ.js → dspy-rlm-engine-US8YsYGq.js} +5 -5
- package/dist/{dspy-rlm-engine-B5T43BkZ.js.map → dspy-rlm-engine-US8YsYGq.js.map} +1 -1
- package/dist/{eval-campaign-GapmPHIA.js → eval-campaign-ha_aWtk7.js} +14 -12
- package/dist/{eval-campaign-GapmPHIA.js.map → eval-campaign-ha_aWtk7.js.map} +1 -1
- package/dist/evaluation.d.ts +9 -10
- package/dist/evaluation.d.ts.map +1 -1
- package/dist/experiment/index.d.ts +29 -29
- package/dist/experiment/index.d.ts.map +1 -1
- package/dist/experiment/index.js +13 -12
- package/dist/experiment/index.js.map +1 -1
- package/dist/{exporters-DEcXbvgO.js → exporters-Coiv-qO2.js} +2 -2
- package/dist/{exporters-DEcXbvgO.js.map → exporters-Coiv-qO2.js.map} +1 -1
- package/dist/{external-optimizer-process-CEXcCKqd.js → external-optimizer-process-KLcw0Vnc.js} +2 -2
- package/dist/{external-optimizer-process-CEXcCKqd.js.map → external-optimizer-process-KLcw0Vnc.js.map} +1 -1
- package/dist/{external-optimizer-subprocess-laM1-UDV.js → external-optimizer-subprocess-CkyUUV7h.js} +4 -4
- package/dist/{external-optimizer-subprocess-laM1-UDV.js.map → external-optimizer-subprocess-CkyUUV7h.js.map} +1 -1
- package/dist/fuzz.d.ts +15 -15
- package/dist/fuzz.d.ts.map +1 -1
- package/dist/{hf-dataset-CRqtSjnR.js → hf-dataset-kjIfIK_e.js} +3 -3
- package/dist/{hf-dataset-CRqtSjnR.js.map → hf-dataset-kjIfIK_e.js.map} +1 -1
- package/dist/hosted/index.d.ts +8 -8
- package/dist/hosted/index.d.ts.map +1 -1
- package/dist/hosted/index.js +1 -1
- package/dist/{index-EkRNsLr4.d.ts → index-D1ZX28rV.d.ts} +3 -3
- package/dist/{index-EkRNsLr4.d.ts.map → index-D1ZX28rV.d.ts.map} +1 -1
- package/dist/{index-D4pC7Tvd2.d.ts → index-D4pC7Tvd.d.ts} +1 -1
- package/dist/index-D4pC7Tvd.d.ts.map +1 -0
- package/dist/{index-DNgf5gyG2.d.ts → index-DNgf5gyG.d.ts} +1 -1
- package/dist/index-DNgf5gyG.d.ts.map +1 -0
- package/dist/{index-DKmrHA02.d.ts → index-giYkbitB.d.ts} +47 -2
- package/dist/index-giYkbitB.d.ts.map +1 -0
- package/dist/index.d.ts +126 -126
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +52 -42
- package/dist/index.js.map +1 -1
- package/dist/{integrity-BFP4kjGj.js → integrity-BvnVARVl.js} +30 -29
- package/dist/{integrity-BFP4kjGj.js.map → integrity-BvnVARVl.js.map} +1 -1
- package/dist/{integrity-BvHANvnp.js → integrity-uwK7XC-8.js} +2 -2
- package/dist/{integrity-BvHANvnp.js.map → integrity-uwK7XC-8.js.map} +1 -1
- package/dist/{internal-BMFSR8Ns.js → internal-BOe3zmjQ.js} +2 -2
- package/dist/{internal-BMFSR8Ns.js.map → internal-BOe3zmjQ.js.map} +1 -1
- package/dist/{jev-review-Cu9xtIu4.js → jev-review-DZEjDzlF.js} +7 -5
- package/dist/{jev-review-Cu9xtIu4.js.map → jev-review-DZEjDzlF.js.map} +1 -1
- package/dist/jev.d.ts +10 -10
- package/dist/jev.d.ts.map +1 -1
- package/dist/jev.js +2 -2
- package/dist/{judge-calibration-DYmaBtJr.js → judge-calibration-C5zgtiJT.js} +2 -2
- package/dist/{judge-calibration-DYmaBtJr.js.map → judge-calibration-C5zgtiJT.js.map} +1 -1
- package/dist/kind-factory-CdPPHtGZ.js.map +1 -1
- package/dist/{llm-client-c_bKHTXz.js → llm-client-Bhqoglar.js} +6 -4
- package/dist/{llm-client-c_bKHTXz.js.map → llm-client-Bhqoglar.js.map} +1 -1
- package/dist/{llm-judge-CL6f3utD.js → llm-judge-C-0P1WiT.js} +26 -21
- package/dist/{llm-judge-CL6f3utD.js.map → llm-judge-C-0P1WiT.js.map} +1 -1
- package/dist/matrix/index.d.ts +1 -1
- package/dist/matrix/index.js +1 -1
- package/dist/{matrix-_NLSWsKM.d.ts → matrix--IZKROxD.d.ts} +2 -2
- package/dist/{matrix-_NLSWsKM.d.ts.map → matrix--IZKROxD.d.ts.map} +1 -1
- package/dist/{matrix-pWinRMPl.js → matrix-DnlHn5x1.js} +8 -6
- package/dist/{matrix-pWinRMPl.js.map → matrix-DnlHn5x1.js.map} +1 -1
- package/dist/meta-eval/index.d.ts +102 -59
- package/dist/meta-eval/index.d.ts.map +1 -1
- package/dist/meta-eval/index.js +89 -25
- package/dist/meta-eval/index.js.map +1 -1
- package/dist/{mint-iCD-W43s.js → mint-vlRckJEX.js} +2 -2
- package/dist/{mint-iCD-W43s.js.map → mint-vlRckJEX.js.map} +1 -1
- package/dist/multishot/golden/index.d.ts +23 -23
- package/dist/multishot/golden/index.d.ts.map +1 -1
- package/dist/multishot/golden/index.js +1 -1
- package/dist/multishot/index.d.ts +9 -9
- package/dist/multishot/index.d.ts.map +1 -1
- package/dist/multishot/index.js +1 -1
- package/dist/openapi.json +1 -1
- package/dist/{otel-spans-CQmSxtcS.js → otel-spans-C8YIJWy1.js} +9 -4
- package/dist/{otel-spans-CQmSxtcS.js.map → otel-spans-C8YIJWy1.js.map} +1 -1
- package/dist/{outcome-store-BXlkwMPR.js → outcome-store-DUcMj-3D.js} +2 -2
- package/dist/{outcome-store-BXlkwMPR.js.map → outcome-store-DUcMj-3D.js.map} +1 -1
- package/dist/package-version-D7lQHt_-.js.map +1 -1
- package/dist/{paired-arms-DKZFVPt8.js → paired-arms-DpS_h8ON.js} +8 -5
- package/dist/{paired-arms-DKZFVPt8.js.map → paired-arms-DpS_h8ON.js.map} +1 -1
- package/dist/{paired-binary-fzkwENI4.js → paired-binary-OMBVRtc-.js} +2 -2
- package/dist/{paired-binary-fzkwENI4.js.map → paired-binary-OMBVRtc-.js.map} +1 -1
- package/dist/{paired-promotion-decision-D5B1Cxc2.js → paired-promotion-decision-DhrmjdV8.js} +5 -4
- package/dist/{paired-promotion-decision-D5B1Cxc2.js.map → paired-promotion-decision-DhrmjdV8.js.map} +1 -1
- package/dist/{paired-tests-C8iCsioC.js → paired-tests-Ccr16gcW.js} +3 -3
- package/dist/{paired-tests-C8iCsioC.js.map → paired-tests-Ccr16gcW.js.map} +1 -1
- package/dist/pipelines/index.d.ts +19 -19
- package/dist/pipelines/index.d.ts.map +1 -1
- package/dist/pipelines/index.js +5 -4
- package/dist/pipelines/index.js.map +1 -1
- package/dist/{power-and-mde-K_CmO3dc.js → power-and-mde-bie6i965.js} +9 -6
- package/dist/{power-and-mde-K_CmO3dc.js.map → power-and-mde-bie6i965.js.map} +1 -1
- package/dist/{produced-state-DZMW8cuF.js → produced-state-CR2tAPLf.js} +8 -5
- package/dist/{produced-state-DZMW8cuF.js.map → produced-state-CR2tAPLf.js.map} +1 -1
- package/dist/{promotion-policy-C6FAJvmZ.js → promotion-policy-BI1RxzwD.js} +4 -4
- package/dist/{promotion-policy-C6FAJvmZ.js.map → promotion-policy-BI1RxzwD.js.map} +1 -1
- package/dist/{raw-provider-sink-CXuOhLvT.js → raw-provider-sink-rB2DYxr6.js} +2 -2
- package/dist/{raw-provider-sink-CXuOhLvT.js.map → raw-provider-sink-rB2DYxr6.js.map} +1 -1
- package/dist/{release-confidence-XQm_xjhR.js → release-confidence-CiFg1EPq.js} +3 -2
- package/dist/{release-confidence-XQm_xjhR.js.map → release-confidence-CiFg1EPq.js.map} +1 -1
- package/dist/{report-command-DKJ_yAIj.js → report-command-DzPpFgjB.js} +20 -18
- package/dist/{report-command-DKJ_yAIj.js.map → report-command-DzPpFgjB.js.map} +1 -1
- package/dist/reporting.js +5 -5
- package/dist/{reward-hacking-D0XwhVWE.js → reward-hacking-C-a__ShZ.js} +5 -3
- package/dist/{reward-hacking-D0XwhVWE.js.map → reward-hacking-C-a__ShZ.js.map} +1 -1
- package/dist/rl.d.ts +186 -186
- package/dist/rl.d.ts.map +1 -1
- package/dist/rl.js +21 -17
- package/dist/rl.js.map +1 -1
- package/dist/rollout/index.d.ts +1 -1
- package/dist/rollout/index.js +5 -5
- package/dist/{rollout-B-XBReTC.js → rollout-BUQixCoc.js} +5 -5
- package/dist/{rollout-B-XBReTC.js.map → rollout-BUQixCoc.js.map} +1 -1
- package/dist/{rubric-predictive-validity-CCK-1B7w.js → rubric-predictive-validity-B0Pg-3I5.js} +3 -3
- package/dist/{rubric-predictive-validity-CCK-1B7w.js.map → rubric-predictive-validity-B0Pg-3I5.js.map} +1 -1
- package/dist/{run-record-tNnlQIrD.js → run-record-BB290Y9A.js} +5 -3
- package/dist/{run-record-tNnlQIrD.js.map → run-record-BB290Y9A.js.map} +1 -1
- package/dist/{schema-C1aaAxTf.js → schema-DAMp77zB.js} +24 -20
- package/dist/{schema-C1aaAxTf.js.map → schema-DAMp77zB.js.map} +1 -1
- package/dist/search/index.d.ts +34 -34
- package/dist/search/index.d.ts.map +1 -1
- package/dist/search/index.js +2 -2
- package/dist/{search-kernel-AVgIjXaI.js → search-kernel-Dpniloks.js} +33 -26
- package/dist/{search-kernel-AVgIjXaI.js.map → search-kernel-Dpniloks.js.map} +1 -1
- package/dist/{search-ledger-CL0-mXQ8.js → search-ledger-CXQPwdWz.js} +4 -3
- package/dist/{search-ledger-CL0-mXQ8.js.map → search-ledger-CXQPwdWz.js.map} +1 -1
- package/dist/{search-shipper-Cs6K4km9.js → search-shipper-DhhofkDX.js} +3 -6
- package/dist/{search-shipper-Cs6K4km9.js.map → search-shipper-DhhofkDX.js.map} +1 -1
- package/dist/{search-summary-DvWk0hb3.js → search-summary-C1nAHgkB.js} +4 -4
- package/dist/{search-summary-DvWk0hb3.js.map → search-summary-C1nAHgkB.js.map} +1 -1
- package/dist/{semantic-concept-judge-2SE_kXGF.js → semantic-concept-judge-BcjYg37z.js} +4 -4
- package/dist/{semantic-concept-judge-2SE_kXGF.js.map → semantic-concept-judge-BcjYg37z.js.map} +1 -1
- package/dist/{sequential-Yz3OSskl.js → sequential-DWn_np6P.js} +2 -3
- package/dist/{sequential-Yz3OSskl.js.map → sequential-DWn_np6P.js.map} +1 -1
- package/dist/{server-D56r1lhp.js → server-B4jMhqQn.js} +2 -2
- package/dist/{server-D56r1lhp.js.map → server-B4jMhqQn.js.map} +1 -1
- package/dist/{skillopt-optimization-method-B2syNik0.js → skillopt-optimization-method-Bls9t-G_.js} +9 -9
- package/dist/{skillopt-optimization-method-B2syNik0.js.map → skillopt-optimization-method-Bls9t-G_.js.map} +1 -1
- package/dist/{statistical-heldout-BAFDCo07.js → statistical-heldout-CNLOvRq_.js} +3 -3
- package/dist/{statistical-heldout-BAFDCo07.js.map → statistical-heldout-CNLOvRq_.js.map} +1 -1
- package/dist/statistical-heldout-HswMvzbP.d.ts.map +1 -1
- package/dist/{store-DNe_Uv1Q.js → store-BhNuLdJw.js} +3 -5
- package/dist/{store-DNe_Uv1Q.js.map → store-BhNuLdJw.js.map} +1 -1
- package/dist/store-C4avFjIi.d.ts.map +1 -1
- package/dist/{store-otlp-BPowgPm6.js → store-otlp-yd9lDziN.js} +2 -2
- package/dist/{store-otlp-BPowgPm6.js.map → store-otlp-yd9lDziN.js.map} +1 -1
- package/dist/{store-tool-spans-BaGakFc2.js → store-tool-spans-Biio4_Nw.js} +4 -4
- package/dist/{store-tool-spans-BaGakFc2.js.map → store-tool-spans-Biio4_Nw.js.map} +1 -1
- package/dist/storyboard/index.d.ts +20 -21
- package/dist/storyboard/index.d.ts.map +1 -1
- package/dist/{student-t-BA-Uy51p.js → student-t-BBZYka0E.js} +4 -3
- package/dist/{student-t-BA-Uy51p.js.map → student-t-BBZYka0E.js.map} +1 -1
- package/dist/{summary-report-BeYnvddy.js → summary-report-DLzptLSA.js} +8 -7
- package/dist/{summary-report-BeYnvddy.js.map → summary-report-DLzptLSA.js.map} +1 -1
- package/dist/supervisor-run/index.d.ts +25 -25
- package/dist/supervisor-run/index.d.ts.map +1 -1
- package/dist/supervisor-run/index.js +2 -2
- package/dist/{tool-waste-CJDfxH45.js → tool-waste-B-u3rzCz.js} +11 -6
- package/dist/{tool-waste-CJDfxH45.js.map → tool-waste-B-u3rzCz.js.map} +1 -1
- package/dist/trace-repair/index.d.ts +96 -96
- package/dist/trace-repair/index.d.ts.map +1 -1
- package/dist/trace-repair/index.js +1 -1
- package/dist/trace-repair/index.js.map +1 -1
- package/dist/traces.d.ts +29 -29
- package/dist/traces.d.ts.map +1 -1
- package/dist/traces.js +9 -7
- package/dist/traces.js.map +1 -1
- package/dist/trajectory-replay/index.d.ts +35 -35
- package/dist/trajectory-replay/index.d.ts.map +1 -1
- package/dist/trajectory-replay/index.js +5 -5
- package/dist/trajectory-replay/index.js.map +1 -1
- package/dist/{tree-Ki_gW7z8.js → tree-D48IJUK0.js} +39 -31
- package/dist/{tree-Ki_gW7z8.js.map → tree-D48IJUK0.js.map} +1 -1
- package/dist/wire/index.d.ts +48 -48
- package/dist/wire/index.d.ts.map +1 -1
- package/dist/wire/index.js +1 -1
- package/docs/evaluator-construction.md +71 -0
- package/package.json +18 -36
- package/dist/campaign-L391bmir.js.map +0 -1
- package/dist/define-agent-eval-1YuvJpOs.d.ts.map +0 -1
- package/dist/index-D4pC7Tvd2.d.ts.map +0 -1
- package/dist/index-DKmrHA02.d.ts.map +0 -1
- package/dist/index-DNgf5gyG2.d.ts.map +0 -1
package/dist/wire/index.d.ts
CHANGED
|
@@ -6,7 +6,7 @@ import { z } from "zod";
|
|
|
6
6
|
import { ServerType } from "@hono/node-server";
|
|
7
7
|
import { Hono } from "hono";
|
|
8
8
|
//#region src/wire/schemas.d.ts
|
|
9
|
-
declare const RubricDimensionSchema: z.ZodObject<{
|
|
9
|
+
export declare const RubricDimensionSchema: z.ZodObject<{
|
|
10
10
|
id: z.ZodString;
|
|
11
11
|
description: z.ZodString;
|
|
12
12
|
weight: z.ZodDefault<z.ZodNumber>;
|
|
@@ -17,7 +17,7 @@ declare const FailureModeSchema: z.ZodObject<{
|
|
|
17
17
|
id: z.ZodString;
|
|
18
18
|
description: z.ZodString;
|
|
19
19
|
}, z.core.$strip>;
|
|
20
|
-
declare const RubricSchema: z.ZodObject<{
|
|
20
|
+
export declare const RubricSchema: z.ZodObject<{
|
|
21
21
|
name: z.ZodString;
|
|
22
22
|
description: z.ZodString;
|
|
23
23
|
systemPrompt: z.ZodString;
|
|
@@ -37,7 +37,7 @@ declare const RubricSchema: z.ZodObject<{
|
|
|
37
37
|
description: z.ZodString;
|
|
38
38
|
}, z.core.$strip>>>;
|
|
39
39
|
}, z.core.$strip>;
|
|
40
|
-
declare const JudgeRequestSchema: z.ZodObject<{
|
|
40
|
+
export declare const JudgeRequestSchema: z.ZodObject<{
|
|
41
41
|
rubricName: z.ZodOptional<z.ZodString>;
|
|
42
42
|
rubric: z.ZodOptional<z.ZodObject<{
|
|
43
43
|
name: z.ZodString;
|
|
@@ -63,7 +63,7 @@ declare const JudgeRequestSchema: z.ZodObject<{
|
|
|
63
63
|
context: z.ZodOptional<z.ZodRecord<z.ZodString, z.ZodUnknown>>;
|
|
64
64
|
model: z.ZodOptional<z.ZodString>;
|
|
65
65
|
}, z.core.$strip>;
|
|
66
|
-
declare const JudgeResultSchema: z.ZodObject<{
|
|
66
|
+
export declare const JudgeResultSchema: z.ZodObject<{
|
|
67
67
|
composite: z.ZodNumber;
|
|
68
68
|
dimensions: z.ZodRecord<z.ZodString, z.ZodNumber>;
|
|
69
69
|
failureModes: z.ZodDefault<z.ZodArray<z.ZodString>>;
|
|
@@ -84,7 +84,7 @@ declare const RubricInfoSchema: z.ZodObject<{
|
|
|
84
84
|
failureModes: z.ZodDefault<z.ZodArray<z.ZodString>>;
|
|
85
85
|
rubricVersion: z.ZodString;
|
|
86
86
|
}, z.core.$strip>;
|
|
87
|
-
declare const ListRubricsResponseSchema: z.ZodObject<{
|
|
87
|
+
export declare const ListRubricsResponseSchema: z.ZodObject<{
|
|
88
88
|
rubrics: z.ZodArray<z.ZodObject<{
|
|
89
89
|
name: z.ZodString;
|
|
90
90
|
description: z.ZodString;
|
|
@@ -97,13 +97,13 @@ declare const ListRubricsResponseSchema: z.ZodObject<{
|
|
|
97
97
|
rubricVersion: z.ZodString;
|
|
98
98
|
}, z.core.$strip>>;
|
|
99
99
|
}, z.core.$strip>;
|
|
100
|
-
declare const VersionResponseSchema: z.ZodObject<{
|
|
100
|
+
export declare const VersionResponseSchema: z.ZodObject<{
|
|
101
101
|
package: z.ZodString;
|
|
102
102
|
version: z.ZodString;
|
|
103
103
|
wireVersion: z.ZodString;
|
|
104
104
|
apiSurface: z.ZodArray<z.ZodString>;
|
|
105
105
|
}, z.core.$strip>;
|
|
106
|
-
declare const HealthResponseSchema: z.ZodObject<{
|
|
106
|
+
export declare const HealthResponseSchema: z.ZodObject<{
|
|
107
107
|
status: z.ZodLiteral<"ok">;
|
|
108
108
|
uptimeSec: z.ZodNumber;
|
|
109
109
|
}, z.core.$strip>;
|
|
@@ -113,7 +113,7 @@ declare const HealthResponseSchema: z.ZodObject<{
|
|
|
113
113
|
* wire schema so non-TypeScript clients can validate without depending
|
|
114
114
|
* on internal types.
|
|
115
115
|
*/
|
|
116
|
-
declare const TraceEventSchema: z.ZodObject<{
|
|
116
|
+
export declare const TraceEventSchema: z.ZodObject<{
|
|
117
117
|
eventId: z.ZodString;
|
|
118
118
|
runId: z.ZodString;
|
|
119
119
|
spanId: z.ZodOptional<z.ZodString>;
|
|
@@ -130,7 +130,7 @@ declare const TraceEventSchema: z.ZodObject<{
|
|
|
130
130
|
timestamp: z.ZodNumber;
|
|
131
131
|
payload: z.ZodRecord<z.ZodString, z.ZodUnknown>;
|
|
132
132
|
}, z.core.$strip>;
|
|
133
|
-
declare const TracesIngestRequestSchema: z.ZodObject<{
|
|
133
|
+
export declare const TracesIngestRequestSchema: z.ZodObject<{
|
|
134
134
|
events: z.ZodArray<z.ZodObject<{
|
|
135
135
|
eventId: z.ZodString;
|
|
136
136
|
runId: z.ZodString;
|
|
@@ -149,7 +149,7 @@ declare const TracesIngestRequestSchema: z.ZodObject<{
|
|
|
149
149
|
payload: z.ZodRecord<z.ZodString, z.ZodUnknown>;
|
|
150
150
|
}, z.core.$strip>>;
|
|
151
151
|
}, z.core.$strip>;
|
|
152
|
-
declare const TracesIngestResponseSchema: z.ZodObject<{
|
|
152
|
+
export declare const TracesIngestResponseSchema: z.ZodObject<{
|
|
153
153
|
accepted: z.ZodNumber;
|
|
154
154
|
rejected: z.ZodNumber;
|
|
155
155
|
errors: z.ZodDefault<z.ZodArray<z.ZodObject<{
|
|
@@ -157,7 +157,7 @@ declare const TracesIngestResponseSchema: z.ZodObject<{
|
|
|
157
157
|
message: z.ZodString;
|
|
158
158
|
}, z.core.$strip>>>;
|
|
159
159
|
}, z.core.$strip>;
|
|
160
|
-
declare const FeedbackTrajectorySchema: z.ZodObject<{
|
|
160
|
+
export declare const FeedbackTrajectorySchema: z.ZodObject<{
|
|
161
161
|
id: z.ZodString;
|
|
162
162
|
projectId: z.ZodOptional<z.ZodString>;
|
|
163
163
|
scenarioId: z.ZodOptional<z.ZodString>;
|
|
@@ -282,38 +282,38 @@ declare const FeedbackTrajectorySchema: z.ZodObject<{
|
|
|
282
282
|
updatedAt: z.ZodOptional<z.ZodString>;
|
|
283
283
|
metadata: z.ZodOptional<z.ZodRecord<z.ZodString, z.ZodUnknown>>;
|
|
284
284
|
}, z.core.$strip>;
|
|
285
|
-
declare const FeedbackIngestResponseSchema: z.ZodObject<{
|
|
285
|
+
export declare const FeedbackIngestResponseSchema: z.ZodObject<{
|
|
286
286
|
id: z.ZodString;
|
|
287
287
|
persisted: z.ZodBoolean;
|
|
288
288
|
}, z.core.$strip>;
|
|
289
|
-
type TraceEvent = z.infer<typeof TraceEventSchema>;
|
|
290
|
-
type TracesIngestRequest = z.infer<typeof TracesIngestRequestSchema>;
|
|
291
|
-
type TracesIngestResponse = z.infer<typeof TracesIngestResponseSchema>;
|
|
292
|
-
type FeedbackTrajectory = z.infer<typeof FeedbackTrajectorySchema>;
|
|
293
|
-
type FeedbackIngestResponse = z.infer<typeof FeedbackIngestResponseSchema>;
|
|
294
|
-
declare const ErrorResponseSchema: z.ZodObject<{
|
|
289
|
+
export type TraceEvent = z.infer<typeof TraceEventSchema>;
|
|
290
|
+
export type TracesIngestRequest = z.infer<typeof TracesIngestRequestSchema>;
|
|
291
|
+
export type TracesIngestResponse = z.infer<typeof TracesIngestResponseSchema>;
|
|
292
|
+
export type FeedbackTrajectory = z.infer<typeof FeedbackTrajectorySchema>;
|
|
293
|
+
export type FeedbackIngestResponse = z.infer<typeof FeedbackIngestResponseSchema>;
|
|
294
|
+
export declare const ErrorResponseSchema: z.ZodObject<{
|
|
295
295
|
error: z.ZodObject<{
|
|
296
296
|
code: z.ZodString;
|
|
297
297
|
message: z.ZodString;
|
|
298
298
|
details: z.ZodOptional<z.ZodUnknown>;
|
|
299
299
|
}, z.core.$strip>;
|
|
300
300
|
}, z.core.$strip>;
|
|
301
|
-
type RubricDimension = z.infer<typeof RubricDimensionSchema>;
|
|
302
|
-
type FailureMode = z.infer<typeof FailureModeSchema>;
|
|
303
|
-
type Rubric = z.infer<typeof RubricSchema>;
|
|
304
|
-
type JudgeRequest = z.infer<typeof JudgeRequestSchema>;
|
|
305
|
-
type JudgeResult = z.infer<typeof JudgeResultSchema>;
|
|
306
|
-
type RubricInfo = z.infer<typeof RubricInfoSchema>;
|
|
307
|
-
type ListRubricsResponse = z.infer<typeof ListRubricsResponseSchema>;
|
|
308
|
-
type VersionResponse = z.infer<typeof VersionResponseSchema>;
|
|
309
|
-
type ErrorResponse = z.infer<typeof ErrorResponseSchema>;
|
|
301
|
+
export type RubricDimension = z.infer<typeof RubricDimensionSchema>;
|
|
302
|
+
export type FailureMode = z.infer<typeof FailureModeSchema>;
|
|
303
|
+
export type Rubric = z.infer<typeof RubricSchema>;
|
|
304
|
+
export type JudgeRequest = z.infer<typeof JudgeRequestSchema>;
|
|
305
|
+
export type JudgeResult = z.infer<typeof JudgeResultSchema>;
|
|
306
|
+
export type RubricInfo = z.infer<typeof RubricInfoSchema>;
|
|
307
|
+
export type ListRubricsResponse = z.infer<typeof ListRubricsResponseSchema>;
|
|
308
|
+
export type VersionResponse = z.infer<typeof VersionResponseSchema>;
|
|
309
|
+
export type ErrorResponse = z.infer<typeof ErrorResponseSchema>;
|
|
310
310
|
/**
|
|
311
311
|
* Bump the major on any breaking change to a request/response schema — the
|
|
312
312
|
* Python client compares majors and refuses a mismatch. Bump the minor when a
|
|
313
313
|
* response value changes meaning without changing its shape, which a client
|
|
314
314
|
* can detect but does not have to act on.
|
|
315
315
|
*/
|
|
316
|
-
declare const WIRE_VERSION = "1.1.0";
|
|
316
|
+
export declare const WIRE_VERSION = "1.1.0";
|
|
317
317
|
/**
|
|
318
318
|
* The digest scheme `hashRubric` mints, named inside the value it produces.
|
|
319
319
|
* A consumer holding a tag from an earlier release reads a different scheme
|
|
@@ -321,7 +321,7 @@ declare const WIRE_VERSION = "1.1.0";
|
|
|
321
321
|
* changed" — the two are indistinguishable when the value is an untagged hex
|
|
322
322
|
* string.
|
|
323
323
|
*/
|
|
324
|
-
declare const RUBRIC_VERSION_SCHEME = "sha256-rfc8785";
|
|
324
|
+
export declare const RUBRIC_VERSION_SCHEME = "sha256-rfc8785";
|
|
325
325
|
/**
|
|
326
326
|
* Stable identity of a rubric, as `<name>@sha256-rfc8785:<hex>`. Two scores
|
|
327
327
|
* are comparable when this matches exactly.
|
|
@@ -331,17 +331,17 @@ declare const RUBRIC_VERSION_SCHEME = "sha256-rfc8785";
|
|
|
331
331
|
* property of the value alone — so the same rubric produces the same tag on
|
|
332
332
|
* every machine, which a locale-collated key order does not guarantee.
|
|
333
333
|
*/
|
|
334
|
-
declare function hashRubric(rubric: Rubric): string;
|
|
334
|
+
export declare function hashRubric(rubric: Rubric): string;
|
|
335
335
|
//#endregion
|
|
336
336
|
//#region src/wire/handlers.d.ts
|
|
337
337
|
/** Caller-fixable error. The transport renders this to 4xx + ErrorResponse. */
|
|
338
|
-
declare class WireError extends Error {
|
|
338
|
+
export declare class WireError extends Error {
|
|
339
339
|
readonly code: string;
|
|
340
340
|
readonly status: number;
|
|
341
341
|
readonly details?: unknown;
|
|
342
342
|
constructor(code: string, message: string, status?: number, details?: unknown);
|
|
343
343
|
}
|
|
344
|
-
interface HandleJudgeOptions {
|
|
344
|
+
export interface HandleJudgeOptions {
|
|
345
345
|
costLedger?: CostLedgerHandle;
|
|
346
346
|
costPhase?: string;
|
|
347
347
|
/**
|
|
@@ -355,9 +355,9 @@ interface HandleJudgeOptions {
|
|
|
355
355
|
pricing?: CustomTokenPricing;
|
|
356
356
|
signal?: AbortSignal;
|
|
357
357
|
}
|
|
358
|
-
declare function handleJudge(req: JudgeRequest, options?: HandleJudgeOptions): Promise<JudgeResult>;
|
|
359
|
-
declare function handleListRubrics(): ListRubricsResponse;
|
|
360
|
-
declare function handleVersion(): VersionResponse;
|
|
358
|
+
export declare function handleJudge(req: JudgeRequest, options?: HandleJudgeOptions): Promise<JudgeResult>;
|
|
359
|
+
export declare function handleListRubrics(): ListRubricsResponse;
|
|
360
|
+
export declare function handleVersion(): VersionResponse;
|
|
361
361
|
/**
|
|
362
362
|
* Pluggable stores the wire layer routes ingestion writes into. Both
|
|
363
363
|
* are optional — when omitted, the corresponding endpoint returns 503.
|
|
@@ -366,7 +366,7 @@ declare function handleVersion(): VersionResponse;
|
|
|
366
366
|
* `FileSystemFeedbackTrajectoryStore` here. Tests substitute in-memory
|
|
367
367
|
* stores.
|
|
368
368
|
*/
|
|
369
|
-
interface IngestionStores {
|
|
369
|
+
export interface IngestionStores {
|
|
370
370
|
traceStore?: TraceStore;
|
|
371
371
|
feedbackStore?: FeedbackTrajectoryStore;
|
|
372
372
|
}
|
|
@@ -380,13 +380,13 @@ interface IngestionStores {
|
|
|
380
380
|
* de-duplicate by `eventId` downstream — production traces frequently
|
|
381
381
|
* land via at-least-once buses (Kafka, SQS) where dedup is unavoidable.
|
|
382
382
|
*/
|
|
383
|
-
declare function handleTracesIngest(req: TracesIngestRequest, stores: IngestionStores): Promise<TracesIngestResponse>;
|
|
383
|
+
export declare function handleTracesIngest(req: TracesIngestRequest, stores: IngestionStores): Promise<TracesIngestResponse>;
|
|
384
384
|
/**
|
|
385
385
|
* `POST /v1/feedback` — accept a single `FeedbackTrajectory` from the
|
|
386
386
|
* production runtime. Idempotent on `id`: re-posting the same trajectory
|
|
387
387
|
* replaces the prior record.
|
|
388
388
|
*/
|
|
389
|
-
declare function handleFeedbackIngest(req: FeedbackTrajectory, stores: IngestionStores): Promise<FeedbackIngestResponse>;
|
|
389
|
+
export declare function handleFeedbackIngest(req: FeedbackTrajectory, stores: IngestionStores): Promise<FeedbackIngestResponse>;
|
|
390
390
|
//#endregion
|
|
391
391
|
//#region node_modules/.pnpm/openapi3-ts@4.6.1/node_modules/openapi3-ts/dist/model/specification-extension.d.ts
|
|
392
392
|
type IExtensionName = `x-${string}`;
|
|
@@ -703,7 +703,7 @@ interface SecurityRequirementObject {
|
|
|
703
703
|
}
|
|
704
704
|
//#endregion
|
|
705
705
|
//#region src/wire/openapi.d.ts
|
|
706
|
-
declare function buildOpenApi(packageVersion: string): OpenAPIObject;
|
|
706
|
+
export declare function buildOpenApi(packageVersion: string): OpenAPIObject;
|
|
707
707
|
//#endregion
|
|
708
708
|
//#region src/wire/rpc.d.ts
|
|
709
709
|
interface RpcRequest {
|
|
@@ -725,17 +725,17 @@ interface RpcOptions {
|
|
|
725
725
|
chat?: ChatClient;
|
|
726
726
|
judgeModel?: string;
|
|
727
727
|
}
|
|
728
|
-
declare function dispatchRpc(req: RpcRequest, options?: RpcOptions): Promise<RpcSuccess | RpcError>;
|
|
728
|
+
export declare function dispatchRpc(req: RpcRequest, options?: RpcOptions): Promise<RpcSuccess | RpcError>;
|
|
729
729
|
/** Read one JSON request from stdin, write one JSON response to stdout. */
|
|
730
|
-
declare function runRpcOnce(method?: string, options?: RpcOptions): Promise<number>;
|
|
730
|
+
export declare function runRpcOnce(method?: string, options?: RpcOptions): Promise<number>;
|
|
731
731
|
/** Read JSONL requests from stdin, write JSONL responses to stdout. */
|
|
732
|
-
declare function runRpcBatch(method?: string, options?: RpcOptions): Promise<number>;
|
|
732
|
+
export declare function runRpcBatch(method?: string, options?: RpcOptions): Promise<number>;
|
|
733
733
|
//#endregion
|
|
734
734
|
//#region src/wire/rubrics.d.ts
|
|
735
735
|
/** Get a built-in rubric by name, or undefined. */
|
|
736
|
-
declare function getBuiltinRubric(name: string): Rubric | undefined;
|
|
736
|
+
export declare function getBuiltinRubric(name: string): Rubric | undefined;
|
|
737
737
|
/** List built-in rubrics with their stable versions. */
|
|
738
|
-
declare function listBuiltinRubrics(): {
|
|
738
|
+
export declare function listBuiltinRubrics(): {
|
|
739
739
|
name: string;
|
|
740
740
|
description: string;
|
|
741
741
|
dimensions: {
|
|
@@ -772,14 +772,14 @@ interface CreateAppOptions {
|
|
|
772
772
|
bearer: string | ((token: string) => boolean | Promise<boolean>);
|
|
773
773
|
};
|
|
774
774
|
}
|
|
775
|
-
declare function createApp(opts?: CreateAppOptions): Hono<import("hono/types").BlankEnv, import("hono/types").BlankSchema, "/">;
|
|
775
|
+
export declare function createApp(opts?: CreateAppOptions): Hono<import("hono/types").BlankEnv, import("hono/types").BlankSchema, "/">;
|
|
776
776
|
interface ServeOptions extends CreateAppOptions {
|
|
777
777
|
/** Default 5005. */
|
|
778
778
|
port?: number;
|
|
779
779
|
/** Default '127.0.0.1'. Set to '0.0.0.0' to listen on all interfaces. */
|
|
780
780
|
host?: string;
|
|
781
781
|
}
|
|
782
|
-
declare function startServer(opts?: ServeOptions): ServerType;
|
|
782
|
+
export declare function startServer(opts?: ServeOptions): ServerType;
|
|
783
783
|
interface StartedServer {
|
|
784
784
|
server: ServerType;
|
|
785
785
|
/** The OS-assigned port. When opts.port was 0, this is the actual port the
|
|
@@ -796,7 +796,7 @@ interface StartedServer {
|
|
|
796
796
|
* listening and surfaces the resolved bound port. Use this from smoke tests
|
|
797
797
|
* (`startServerAsync({ port: 0 })`) and any caller that needs to dial back.
|
|
798
798
|
*/
|
|
799
|
-
declare function startServerAsync(opts?: ServeOptions): Promise<StartedServer>;
|
|
799
|
+
export declare function startServerAsync(opts?: ServeOptions): Promise<StartedServer>;
|
|
800
800
|
//#endregion
|
|
801
|
-
export {
|
|
801
|
+
export type { RpcOptions, ServeOptions, StartedServer };
|
|
802
802
|
//# sourceMappingURL=index.d.ts.map
|
package/dist/wire/index.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","names":[],"sources":["../../src/wire/schemas.ts","../../src/wire/handlers.ts","../../node_modules/.pnpm/openapi3-ts@4.6.1/node_modules/openapi3-ts/dist/model/specification-extension.d.ts","../../node_modules/.pnpm/openapi3-ts@4.6.1/node_modules/openapi3-ts/dist/model/oas-common.d.ts","../../node_modules/.pnpm/openapi3-ts@4.6.1/node_modules/openapi3-ts/dist/model/openapi31.d.ts","../../src/wire/openapi.ts","../../src/wire/rpc.ts","../../src/wire/rubrics.ts","../../src/wire/server.ts"],"x_google_ignoreList":[2,3,4],"mappings":";;;;;;;;
|
|
1
|
+
{"version":3,"file":"index.d.ts","names":["WireFeedbackTrajectory"],"sources":["../../src/wire/schemas.ts","../../src/wire/handlers.ts","../../node_modules/.pnpm/openapi3-ts@4.6.1/node_modules/openapi3-ts/dist/model/specification-extension.d.ts","../../node_modules/.pnpm/openapi3-ts@4.6.1/node_modules/openapi3-ts/dist/model/oas-common.d.ts","../../node_modules/.pnpm/openapi3-ts@4.6.1/node_modules/openapi3-ts/dist/model/openapi31.d.ts","../../src/wire/openapi.ts","../../src/wire/rpc.ts","../../src/wire/rubrics.ts","../../src/wire/server.ts"],"x_google_ignoreList":[2,3,4],"mappings":";;;;;;;;qBAoBa,uBAAqB,EAAA;;;;;;GAkBL,EAAA,KAAA;cAEvB,mBAAiB,EAAA;;;GAKE,EAAA,KAAA;qBAIZ,cAAY,EAAA;;;;;;;;;;;;;;;;;;;GA6BL,EAAA,KAAA;qBAIP,oBAAkB,EAAA;;;;;;;;;;;;;;;;;;;;;;;;;GAwBL,EAAA,KAAA;qBAEb,mBAAiB,EAAA;;;;;;;;;GA6BL,EAAA,KAAA;cAInB,kBAAgB,EAAA;;;;;;;;;;GAcE,EAAA,KAAA;qBAEX,2BAAyB,EAAA;;;;;;;;;;;;GAIL,EAAA,KAAA;qBAIpB,uBAAqB,EAAA;;;;;GAWL,EAAA,KAAA;qBAEhB,sBAAoB,EAAA;;;GAKL,EAAA,KAAA;;;;;;;qBAUf,kBAAgB,EAAA;;;;;;;;;;;;;;;;GA0BL,EAAA,KAAA;qBAEX,2BAAyB,EAAA;;;;;;;;;;;;;;;;;;GAQL,EAAA,KAAA;qBAEpB,4BAA0B,EAAA;;;;;;;GAiBL,EAAA,KAAA;qBA2DrB,0BAAwB,EAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BL,EAAA,KAAA;qBAEnB,8BAA4B,EAAA;;;GAKL,EAAA,KAAA;YAExB,aAAa,EAAE,aAAa;YAC5B,sBAAsB,EAAE,aAAa;YACrC,uBAAuB,EAAE,aAAa;YACtC,qBAAqB,EAAE,aAAa;YACpC,yBAAyB,EAAE,aAAa;qBAIvC,qBAAmB,EAAA;;;;;;GAcL,EAAA,KAAA;YAIf,kBAAkB,EAAE,aAAa;YACjC,cAAc,EAAE,aAAa;YAC7B,SAAS,EAAE,aAAa;YACxB,eAAe,EAAE,aAAa;YAC9B,cAAc,EAAE,aAAa;YAC7B,aAAa,EAAE,aAAa;YAC5B,sBAAsB,EAAE,aAAa;YACrC,kBAAkB,EAAE,aAAa;YACjC,gBAAgB,EAAE,aAAa;;;;;;;qBAU9B;;;;;;;;qBASA;;;;;;;;;;wBAWG,WAAW,QAAQ;;;;qBCpXtB,kBAAkB;WAEX;WAEA;WACA;EAJlB,YACkB,cAChB,iBACgB,iBACA;;iBAgJH;EACf,aAAa;EACb;;;;;;EAMA,OAAO;EACP;;EAEA,UAAU;EACV,SAAS;;wBAGW,YACpB,KAAK,cACL,UAAS,qBACR,QAAQ;wBAuEK,qBAAqB;wBAMrB,iBAAiB;;;;;;;;;iBAmBhB;EACf,aAAa;EACb,gBAAgB;;;;;;;;;;;;wBAaI,mBACpB,KAAK,qBACL,QAAQ,kBACP,QAAQ;;;;;;wBA8BW,qBACpB,KAAKA,oBACL,QAAQ,kBACP,QAAQ;;;KC9VC;KACA;KACA;GACP,eAAe,iBAAiB;;;;UCFpB,qBAAqB;EAClC;EACA;EACA;KACK,YAAY;;;UAGJ,6BAA6B;EAC1C;EACA;EACA;;;;UCPa,sBAAsB;EACnC;EACA,MAAM;EACN,UAAU;EACV,QAAQ;EACR,aAAa;EACb,WAAW;EACX,OAAO;EACP,eAAe;EACf,WAAW;;UAEE,mBAAmB;EAChC;EACA;EACA;EACA,UAAU;EACV,UAAU;EACV;;UAEa,sBAAsB;EACnC;EACA;EACA;;UAEa,sBAAsB;EACnC;EACA;EACA;;UAEa,yBAAyB;EACtC;KACK,iBAAiB,eAAe;;EAErC;KACK,mBAAmB,iBAAiB;;EAEzC;KACK,oBAAoB,kBAAkB;;EAE3C;KACK,kBAAkB,gBAAgB;;EAEvC;KACK,kBAAkB,oBAAoB;;EAE3C;KACK,iBAAiB,eAAe;;EAErC;KACK,yBAAyB,uBAAuB;;EAErD;KACK,eAAe,aAAa;;EAEjC;KACK,mBAAmB,iBAAiB;;EAEzC;KACK,mBAAmB,iBAAiB;;;UAG5B,oBAAoB;GAChC,eAAe;;UAIH,uBAAuB;EACpC;EACA;EACA;EACA,MAAM;EACN,MAAM;EACN,OAAO;EACP,SAAS;EACT,UAAU;EACV,OAAO;EACP,QAAQ;EACR,QAAQ;EACR,UAAU;EACV,cAAc,kBAAkB;;UAEnB,wBAAwB;EACrC;EACA;EACA;EACA,eAAe;EACf;EACA,cAAc,kBAAkB;EAChC,cAAc,oBAAoB;EAClC,YAAY;EACZ,YAAY;EACZ;EACA,WAAW;EACX,UAAU;;UAEG,oCAAoC;EACjD;EACA;;KAEQ;KACA;UACK,4BAA4B;EACzC;EACA;EACA;EACA;EACA,QAAQ;EACR;EACA;EACA,SAAS,eAAe;EACxB;KACK,gBAAgB,gBAAgB;;EAErC;EACA,UAAU;;UAEG,wBAAwB;EACrC;EACA,IAAI;;UAES,0BAA0B;EACvC;EACA,SAAS;EACT;;UAEa;GACZ,oBAAoB;;UAER,wBAAwB;EACrC,SAAS,eAAe;EACxB,WAAW;EACX;EACA,WAAW;;UAEE,uBAAuB;GACnC,mBAAmB;;UAEP;EACb;EACA;KACK,cAAc,eAAe;;EAElC;EACA;EACA;GACC;;UAEY,wBAAwB;EACrC,UAAU,iBAAiB;GAC1B,qBAAqB,iBAAiB;;UAE1B,uBAAuB;EACpC;EACA,UAAU;EACV,UAAU;EACV,QAAQ;;UAEK,wBAAwB;GACpC,eAAe,iBAAiB;;UAEpB,uBAAuB;GACnC,eAAe;;UAEH;GACZ,eAAe,eAAe;;UAElB;EACb;EACA;EACA;EACA;GACC;;UAEY;GACZ,eAAe,aAAa;;UAEhB,mBAAmB;EAChC;EACA;EACA,aAAa;EACb;EACA;EACA,SAAS;GACR;;UAEY;GACZ;;UAEY,qBAAqB;EAClC;;UAEa,kBAAkB;EAC/B;EACA;EACA,eAAe;GACd;;UAEY;GACZ,eAAe,gBAAgB;;UAEnB;EACb;EACA;EACA;;KAGQ;UACK,qBAAqB;EAClC;EACA,gBAAgB;EAChB;EACA;EACA,MAAM;EACN,eAAe;EACf;EACA;EACA;EACA,OAAO,mBAAmB;EAC1B;EACA,SAAS,eAAe;EACxB,SAAS,eAAe;EACxB,SAAS,eAAe;EACxB,MAAM,eAAe;EACrB,QAAQ,eAAe;EACvB;KACK,uBAAuB,eAAe;;EAE3C,uBAAuB,eAAe;EACtC,gBAAgB,eAAe;EAC/B;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA,eAAe,eAAe;EAC9B;EACA;;UAMa;EACb;EACA;KACK;;;UAGQ,kBAAkB;EAC/B;EACA;EACA;EACA;EACA;;KAEQ;UACK,6BAA6B;EAC1C,MAAM;EACN;EACA;EACA;EACA;EACA;EACA,QAAQ;EACR;;UAEa,yBAAyB;EACtC,WAAW;EACX,WAAW;EACX,oBAAoB;EACpB,oBAAoB;;UAEP,wBAAwB;EACrC;EACA;EACA;EACA,QAAQ;;UAEK,qBAAqB;GACjC;;UAEY;GACZ;;;;wBChRW,aAAa,yBAAyB;;;UCX5C;EACR;EACA;;UAGQ;EACR;;UAGQ;EACR;IAAS;IAAc;IAAiB;;;UAGzB;;EAEf,OAAO;EACP;;wBAGoB,YACpB,KAAK,YACL,UAAS,aACR,QAAQ,aAAa;;wBAqDF,WAAW,iBAAiB,UAAS,aAAkB;;wBAuBvD,YAAY,iBAAiB,UAAS,aAAkB;;;;wBCjB9D,iBAAiB,eAAe;;wBAKhC;;;;;;;;;;;;;UCtEC;;EAEf,SAAS;;;;;;EAMT,OAAO;;EAEP;;;;;;;;;;EAUA;IACE,mBAAmB,4BAA4B;;;wBAMnC,UAAU,OAAM,mBAAqB,0BAAA,+BAAA;UA0IpC,qBAAqB;;EAEpC;;EAEA;;wBAGc,YAAY,OAAM,eAAoB;UAUrC;EACf,QAAQ;;;;EAIR;;EAEA;;EAEA,SAAS;;;;;;;wBAQK,iBAAiB,OAAM,eAAoB,QAAQ"}
|
package/dist/wire/index.js
CHANGED
|
@@ -1,2 +1,2 @@
|
|
|
1
|
-
import { A as WIRE_VERSION, C as RUBRIC_VERSION_SCHEME, D as TracesIngestRequestSchema, E as TraceEventSchema, O as TracesIngestResponseSchema, S as ListRubricsResponseSchema, T as RubricSchema, _ as FeedbackIngestResponseSchema, a as runRpcBatch, b as JudgeRequestSchema, c as WireError, d as handleListRubrics, f as handleTracesIngest, g as ErrorResponseSchema, h as listBuiltinRubrics, i as dispatchRpc, j as hashRubric, k as VersionResponseSchema, l as handleFeedbackIngest, m as getBuiltinRubric, n as startServer, o as runRpcOnce, p as handleVersion, r as startServerAsync, s as buildOpenApi, t as createApp, u as handleJudge, v as FeedbackTrajectorySchema, w as RubricDimensionSchema, x as JudgeResultSchema, y as HealthResponseSchema } from "../server-
|
|
1
|
+
import { A as WIRE_VERSION, C as RUBRIC_VERSION_SCHEME, D as TracesIngestRequestSchema, E as TraceEventSchema, O as TracesIngestResponseSchema, S as ListRubricsResponseSchema, T as RubricSchema, _ as FeedbackIngestResponseSchema, a as runRpcBatch, b as JudgeRequestSchema, c as WireError, d as handleListRubrics, f as handleTracesIngest, g as ErrorResponseSchema, h as listBuiltinRubrics, i as dispatchRpc, j as hashRubric, k as VersionResponseSchema, l as handleFeedbackIngest, m as getBuiltinRubric, n as startServer, o as runRpcOnce, p as handleVersion, r as startServerAsync, s as buildOpenApi, t as createApp, u as handleJudge, v as FeedbackTrajectorySchema, w as RubricDimensionSchema, x as JudgeResultSchema, y as HealthResponseSchema } from "../server-B4jMhqQn.js";
|
|
2
2
|
export { ErrorResponseSchema, FeedbackIngestResponseSchema, FeedbackTrajectorySchema, HealthResponseSchema, JudgeRequestSchema, JudgeResultSchema, ListRubricsResponseSchema, RUBRIC_VERSION_SCHEME, RubricDimensionSchema, RubricSchema, TraceEventSchema, TracesIngestRequestSchema, TracesIngestResponseSchema, VersionResponseSchema, WIRE_VERSION, WireError, buildOpenApi, createApp, dispatchRpc, getBuiltinRubric, handleFeedbackIngest, handleJudge, handleListRubrics, handleTracesIngest, handleVersion, hashRubric, listBuiltinRubrics, runRpcBatch, runRpcOnce, startServer, startServerAsync };
|
|
@@ -0,0 +1,71 @@
|
|
|
1
|
+
# Evaluator construction: plan, execute, audit
|
|
2
|
+
|
|
3
|
+
Use existing Runtime execution and Eval measurements. A product integration should not implement another error-bound formula, check counter, or per-call intent/result protocol.
|
|
4
|
+
|
|
5
|
+
## Plan before spending
|
|
6
|
+
|
|
7
|
+
```ts
|
|
8
|
+
import { planEvaluatorAudit } from '@tangle-network/agent-eval/meta-eval'
|
|
9
|
+
|
|
10
|
+
const policy = {
|
|
11
|
+
confidence: 0.95,
|
|
12
|
+
maxFalseAcceptanceRate: 0.1,
|
|
13
|
+
maxFalseRejectionRate: 0.1,
|
|
14
|
+
}
|
|
15
|
+
const requirements = planEvaluatorAudit({ policy })
|
|
16
|
+
// minimumIndependentUnits: 42 per class for this policy, even with zero errors.
|
|
17
|
+
|
|
18
|
+
const capacity = planEvaluatorAudit({ policy, controls })
|
|
19
|
+
if (!capacity.sufficient) {
|
|
20
|
+
// Collect enough authorized, independent evidence before constructing a checker.
|
|
21
|
+
// Each class reports its actual count and additionalIndependentUnits.
|
|
22
|
+
}
|
|
23
|
+
```
|
|
24
|
+
|
|
25
|
+
A control has `id`, `independentUnitId`, `expected: 'accept' | 'reject'`, and `exposure: 'fresh' | 'development'`. It has no observed judgment. Repeated variants share a source identity. Any development exposure excludes that entire source unit, exactly as in admission.
|
|
26
|
+
|
|
27
|
+
This is a best-case eligibility floor, not power, permission to spend, or an admission certificate. A zero error-probability limit has no finite sample solution. `auditEvaluator` and `auditProbabilityPolicy` still require actual independent evidence. Their existing report and digest contracts are unchanged.
|
|
28
|
+
|
|
29
|
+
The existing `evaluatorAdmissionPolicySchema` and `evaluatorAuditObservationSchema` are exported from `/meta-eval`. Derive consumer contracts and types from these instead of copying policy fields or decision enums.
|
|
30
|
+
|
|
31
|
+
## Retain external check calls
|
|
32
|
+
|
|
33
|
+
```ts
|
|
34
|
+
import { createRetainedDispatch } from '@tangle-network/agent-eval/campaign'
|
|
35
|
+
|
|
36
|
+
const checks = createRetainedDispatch({
|
|
37
|
+
runDir, // one private, durable logical episode directory
|
|
38
|
+
scope: frozenManifest, // authority, code/decoder revision, environment and inputs
|
|
39
|
+
limits: { development: 20, audit: 84 },
|
|
40
|
+
parse: validateRuntimeResult,
|
|
41
|
+
})
|
|
42
|
+
|
|
43
|
+
const result = await checks.run({
|
|
44
|
+
lane: 'audit',
|
|
45
|
+
input: { evaluatorDigest, caseId, inputDigest },
|
|
46
|
+
signal,
|
|
47
|
+
dispatch: () => executeIsolatedCheck(),
|
|
48
|
+
})
|
|
49
|
+
```
|
|
50
|
+
|
|
51
|
+
The limits above illustrate disjoint allowances; choose them from the approved experiment budget and required roster. Development cannot consume the audit reserve. Counts include uncertain calls. They are not dollars and do not replace the existing spend ledger or provider-enforced ceilings.
|
|
52
|
+
|
|
53
|
+
`result.succeeded` means an external-call result was retained, not that the task or checker passed. The value may itself record an execution failure. A settled result replays with `replayed: true`; it is parsed and its digest checked on every read. A result retained before a settlement lock conflict is reconciled by dispatch ID on retry, without repeating the external call. Concurrent callers on one handle share an in-flight call. Independent processes use the existing `CampaignStorage.append` compare-and-swap.
|
|
54
|
+
|
|
55
|
+
A rejected dispatch or cancelled in-flight call returns `outcome_unknown` with its diagnostic and keeps the allowance committed. A process crash or result loss after intent also leaves `outcome_unknown`; a new caller does not repeat it or refund its allowance. A corrupt/missing settled receipt, changed scope, changed limits, or torn history is an error, not a fresh run. Keep the same directory on retry; changing the directory is a new episode, so the existing product job/authority owner must prevent that from becoming a budget reset.
|
|
56
|
+
|
|
57
|
+
The filesystem adapter fsyncs through the existing storage primitive. An in-memory adapter remains in-memory and cannot establish reboot recovery. Caller-owned storage must provide the documented atomic append contract. Keep the directory and result files outside generated code's permissions; content hashes are integrity checks, not a substitute for tenant authorization or a trusted storage boundary.
|
|
58
|
+
|
|
59
|
+
`runEval` still owns scheduling, judges, concurrency and campaign reports. Wrap only the external dispatch that needs uncertain-result protection. The primitive does not introduce a scheduler, database, execution backend or inferred price. Reference and generated-checker execution use the same mechanism.
|
|
60
|
+
|
|
61
|
+
The caller binds every behavior-affecting dependency and decoder revision into scope, validates the complete JSON result, and captures request state before dispatch. Noncanonical or non-round-tripping result encodings are refused rather than silently coerced.
|
|
62
|
+
|
|
63
|
+
## Prove the persistence path
|
|
64
|
+
|
|
65
|
+
```sh
|
|
66
|
+
pnpm exec tsx scripts/prove-retained-dispatch.ts
|
|
67
|
+
```
|
|
68
|
+
|
|
69
|
+
This executes real child processes against filesystem storage, kills an owner after its external write, and checks contention, completed replay, unknown-outcome retention, settlement-lock recovery, independent audit allowance, scope changes, cancellation, receipt identity and corrupted evidence. The maintained test suite runs the same proof. It makes no model calls and is not a semantic evaluator-quality or customer-savings benchmark.
|
|
70
|
+
|
|
71
|
+
An integration still needs its real Runtime/Sandbox execution, independent audit authorities, approved fresh evidence, complete cost reconciliation, and controlled activation proof. This API does not manufacture any of them.
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@tangle-network/agent-eval",
|
|
3
|
-
"version": "0.
|
|
3
|
+
"version": "0.202.0",
|
|
4
4
|
"description": "Evaluate and improve AI agents from runs, traces, judges, and feedback. Compare candidates, cluster failures, measure lift, and gate releases.",
|
|
5
5
|
"homepage": "https://github.com/tangle-network/agent-eval#readme",
|
|
6
6
|
"repository": {
|
|
@@ -224,53 +224,35 @@
|
|
|
224
224
|
"dependencies": {
|
|
225
225
|
"@asteasolutions/zod-to-openapi": "^9.1.0",
|
|
226
226
|
"@hono/node-server": "^2.1.1",
|
|
227
|
-
"@tangle-network/agent-core": "^0.
|
|
228
|
-
"@tangle-network/agent-interface": "^2.
|
|
229
|
-
"@tangle-network/agent-trace-contract": "^
|
|
230
|
-
"hono": "^4.13.
|
|
227
|
+
"@tangle-network/agent-core": "^0.10.2",
|
|
228
|
+
"@tangle-network/agent-interface": "^2.13.1",
|
|
229
|
+
"@tangle-network/agent-trace-contract": "^2.0.0",
|
|
230
|
+
"hono": "^4.13.9",
|
|
231
231
|
"linear-sum-assignment": "1.0.9",
|
|
232
232
|
"re2js": "2.8.6",
|
|
233
|
-
"smol-toml": "1.
|
|
234
|
-
"zod": "4.5
|
|
233
|
+
"smol-toml": "1.9.0",
|
|
234
|
+
"zod": "4.6.5"
|
|
235
235
|
},
|
|
236
236
|
"devDependencies": {
|
|
237
237
|
"@arethetypeswrong/cli": "^0.18.5",
|
|
238
|
-
"@arizeai/phoenix-evals": "^2.
|
|
239
|
-
"@biomejs/biome": "^2.5.
|
|
240
|
-
"@types/node": "^26.
|
|
238
|
+
"@arizeai/phoenix-evals": "^2.6.0",
|
|
239
|
+
"@biomejs/biome": "^2.5.14",
|
|
240
|
+
"@types/node": "^26.6.2",
|
|
241
241
|
"ajv": "8.20.0",
|
|
242
242
|
"autoevals": "^0.3.0",
|
|
243
243
|
"esbuild": "^0.28.2",
|
|
244
|
-
"fast-check": "^4.
|
|
244
|
+
"fast-check": "^4.10.2",
|
|
245
245
|
"husky": "^9.1.7",
|
|
246
246
|
"lib-r-math.js": "^3.0.2",
|
|
247
|
-
"lint-staged": "^17.
|
|
247
|
+
"lint-staged": "^17.5.1",
|
|
248
248
|
"openapi3-ts": "^4.6.1",
|
|
249
|
-
"oxc-parser": "^0.
|
|
249
|
+
"oxc-parser": "^0.151.0",
|
|
250
250
|
"publint": "^0.3.24",
|
|
251
|
-
"tsdown": "^0.
|
|
252
|
-
"tsx": "^4.23.
|
|
251
|
+
"tsdown": "^0.23.0",
|
|
252
|
+
"tsx": "^4.23.15",
|
|
253
253
|
"typescript": "^7.0.2",
|
|
254
|
-
"vitest": "^
|
|
255
|
-
"yaml": "2.9.
|
|
256
|
-
},
|
|
257
|
-
"pnpm": {
|
|
258
|
-
"minimumReleaseAge": 4320,
|
|
259
|
-
"minimumReleaseAgeExclude": [
|
|
260
|
-
"@tangle-network/*",
|
|
261
|
-
"esbuild",
|
|
262
|
-
"vite",
|
|
263
|
-
"zod@4.5.4"
|
|
264
|
-
],
|
|
265
|
-
"overrides": {
|
|
266
|
-
"@arizeai/openinference-core>@opentelemetry/core": "^2.10.0",
|
|
267
|
-
"esbuild@>=0.27.3 <0.28.1": "^0.28.1",
|
|
268
|
-
"fast-uri": "3.1.6",
|
|
269
|
-
"js-yaml": "4.3.1",
|
|
270
|
-
"postcss@<8.5.18": "^8.5.18",
|
|
271
|
-
"vite@>=7.0.0 <=7.3.4": "^7.3.5",
|
|
272
|
-
"ws@>=8.0.0 <8.21.0": "^8.21.0"
|
|
273
|
-
}
|
|
254
|
+
"vitest": "^5.0.1",
|
|
255
|
+
"yaml": "2.9.1"
|
|
274
256
|
},
|
|
275
257
|
"engines": {
|
|
276
258
|
"node": ">=20.19.0"
|
|
@@ -284,5 +266,5 @@
|
|
|
284
266
|
]
|
|
285
267
|
},
|
|
286
268
|
"license": "MIT",
|
|
287
|
-
"packageManager": "pnpm@
|
|
269
|
+
"packageManager": "pnpm@12.6.0"
|
|
288
270
|
}
|