npm - @tangle-network/agent-runtime - Versions diffs - 0.107.2 → 0.107.3 - Mend

@tangle-network/agent-runtime 0.107.2 → 0.107.3

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.

Files changed (4) hide show

package/dist/testing.d.ts CHANGED Viewed

@@ -1,7 +1,20 @@
-import { AgentImprovementProposal } from "@tangle-network/agent-interface";
+import { AgentImprovementProposal, AgentProfile, AgentProfileImprovementMeasuredComparison, SandboxSizePreset } from "@tangle-network/agent-interface";
 //#region src/testing/index.d.ts
+/** A proposal produced by Runtime's opaque profile-improvement path. */
+type AgentProfileImprovementProposalFixture = Omit<AgentImprovementProposal, 'evaluation'> & {
+  evaluation: AgentProfileImprovementMeasuredComparison;
+};
+/** Complete private state for exercising profile activation and restore in consumer tests. */
+interface AgentProfileImprovementFixture {
+  proposal: AgentProfileImprovementProposalFixture;
+  baselineProfile: AgentProfile;
+  candidateProfile: AgentProfile;
+  recommendedSize: SandboxSizePreset;
+}
 /** Load an isolated, production-validated Runtime proposal for consumer tests. */
 declare function loadAgentImprovementProposalFixture(): AgentImprovementProposal;
+/** Load an isolated profile proposal and its private activation state for consumer tests. */
+declare function loadAgentProfileImprovementFixture(): AgentProfileImprovementFixture;
 //#endregion
-export { loadAgentImprovementProposalFixture };
+export { AgentProfileImprovementFixture, AgentProfileImprovementProposalFixture, loadAgentImprovementProposalFixture, loadAgentProfileImprovementFixture };
 //# sourceMappingURL=testing.d.ts.map

package/dist/testing.js CHANGED Viewed

@@ -1,9 +1,10 @@
+import { $ as parseExactAgentProfile, K as applyExactAgentProfileDiff, rt as canonicalCandidateDigest$1, st as immutableCandidateValue } from "./prepare-_WTTffkz.js";
 import { f as verifyAgentImprovementProposal } from "./improvement-cycle-Ulvvbg5p.js";
-//#endregion
-//#region src/testing/index.ts
-const serializedAgentImprovementProposalFixture = JSON.stringify({
+import { SANDBOX_SIZE_PRESET_NAMES } from "@tangle-network/agent-interface";
+//#region src/testing/fixtures/agent-improvement-proposal.json
+var agent_improvement_proposal_default = {
 	changedSurfaces: ["prompt"],
-	digest: "sha256:4827b89fc44ae848377d620839451748615aff0247c0e0c75f889182eeb413b6",
+	digest: "sha256:f309bdca66c941bae76f9cd71d630e3e73419ab0ea53a520c1e775c799c23642",
 	evaluation: {
 		"decision": {
 			"contributingChecks": [
@@ -2363,7 +2364,7 @@ const serializedAgentImprovementProposalFixture = JSON.stringify({
 		],
 		"metadata": {
 			"fixture": "agent-improvement-proposal",
-			"runtimeVersion": "0.107.2"
+			"runtimeVersion": "0.107.3"
 		},
 		"objectives": [
 			{
@@ -2474,8 +2475,8 @@ const serializedAgentImprovementProposalFixture = JSON.stringify({
 			"baselineContentHash": "sha256:5c21ee53e513fc604cb09754e21c392b24a424da0ef37dbf8f1ee4a8a0b08f09",
 			"candidateContentHash": "sha256:60fcbb1c728194bd51d7d19cb732d1c3f1881dce7e0a6266b41c8b98cfd65693",
 			"kind": "agent-eval-loop",
-			"recordDigest": "sha256:6865263c72b0d5b11db601d815e5ef0e7c55d1e4e01b51a335dc9f5a9753278f",
-			"runId": "agent-runtime-0.107.2-proposal-fixture",
+			"recordDigest": "sha256:d8b668e26ebe5fe95570ea35968048227bcb815e9d7593d41c62d995be0ac7d0",
+			"runId": "agent-runtime-0.107.3-proposal-fixture",
 			"schema": "agent-candidate-experiment"
 		}
 	},
@@ -2497,13 +2498,1095 @@ const serializedAgentImprovementProposalFixture = JSON.stringify({
 	}],
 	kind: "agent-improvement-proposal",
 	proposedAt: "2026-07-10T01:00:00.000Z",
-	runId: "agent-runtime-0.107.2-proposal-fixture"
-});
+	runId: "agent-runtime-0.107.3-proposal-fixture"
+};
+//#endregion
+//#region src/testing/fixtures/agent-profile-improvement-proposal.json
+var agent_profile_improvement_proposal_default = {
+	changedSurfaces: ["prompt", "skills"],
+	digest: "sha256:ce7b2603fcc264631f65df4a13433d371f88bf5d96882d62fe691fac8b867cc4",
+	evaluation: {
+		"decision": {
+			"contributingChecks": [{
+				"name": "paired-significance",
+				"passed": true
+			}],
+			"outcome": "ship",
+			"reasons": ["paired comparison passed"]
+		},
+		"diff": "prompt: add source and uncertainty instructions",
+		"evaluation": {
+			"generationsExplored": 1,
+			"measurement": {
+				"cost": {
+					"provenance": "observed",
+					"usd": 66e-8
+				},
+				"wallDurationMs": 0,
+				"workDurationMs": 660
+			},
+			"preparation": {
+				"cost": {
+					"provenance": "observed",
+					"usd": 0
+				},
+				"wallDurationMs": 0
+			},
+			"total": {
+				"cost": {
+					"provenance": "observed",
+					"usd": 66e-8
+				},
+				"wallDurationMs": 0
+			}
+		},
+		"experiment": {
+			"baseline": { "stateDigest": "sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704" },
+			"benchmark": {
+				"suite": {
+					"digest": "sha256:b625b9c8791b5482f5b4e7c2471eb3ca1b719b5424e86d2bcb1044defbba635e",
+					"digestAlgorithm": "rfc8785-sha256",
+					"kind": "agent-profile-improvement-suite",
+					"reps": 3,
+					"seeds": [
+						11,
+						12,
+						13
+					],
+					"splitDigest": "sha256:a8f4bdc4df6d527e21bcfeedf9fcbc007723329c25da5da8c9ad3e6887623e22",
+					"taskDigests": ["sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e"]
+				},
+				"tasks": [{
+					"digest": "sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e",
+					"digestAlgorithm": "rfc8785-sha256",
+					"grader": {
+						"artifact": {
+							"byteLength": 1,
+							"locator": {
+								"bucket": "agent-eval",
+								"key": "graders/profile-quality.json",
+								"kind": "s3",
+								"region": "us-east-1"
+							},
+							"sha256": "sha256:1f576a8a9d7f4af7bf070698b2be73904847d6f28efec0c2366d1f2bc6117e0f"
+						},
+						"format": "tangle-grader",
+						"name": "profile-quality",
+						"version": "1"
+					},
+					"kind": "agent-profile-improvement-task",
+					"limits": {
+						"maxCostUsd": 1,
+						"maxInputTokens": 1e3,
+						"maxModelCalls": 2,
+						"maxOutputTokens": 1e3,
+						"maxSteps": 10,
+						"timeoutMs": 3e4
+					},
+					"model": {
+						"model": "claude-sonnet-4-6",
+						"provider": "anthropic",
+						"reasoningEffort": "medium",
+						"requested": "anthropic/claude-sonnet-4-6",
+						"snapshot": "2026-06-01"
+					},
+					"scenario": {
+						"digest": "sha256:b50dec7031d699e2e0cfca8b4fd1eb958b687030a1af7a5339090dc9ca0892bf",
+						"id": "support-case-1",
+						"kind": "support-case"
+					}
+				}]
+			},
+			"candidate": { "stateDigest": "sha256:103f77bc8481601eef1ad5fe6ba84a40dffabc3a44f421f8c8559121edab84e9" },
+			"candidateLineage": {
+				"developmentSplitDigest": "sha256:02da010d5b66a91bc293d928ed8297f05950f56f3127f058f94b8766bc6a2b76",
+				"parentDigests": ["sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704"],
+				"runIds": ["profile-improvement-1"],
+				"source": "optimizer"
+			},
+			"change": [{
+				"id": "add-source-and-uncertainty",
+				"kind": "agent-profile-diff",
+				"set": {
+					"prompt": { "systemPrompt": "Answer directly, cite the source, and state uncertainty." },
+					"resources": { "skills": [{
+						"content": "Cite the evidence you use.",
+						"kind": "inline",
+						"name": "sources.SKILL.md"
+					}] }
+				},
+				"source": {
+					"artifacts": ["traces://run/profile-improvement-1"],
+					"kind": "optimizer"
+				}
+			}],
+			"digest": "sha256:ae081dce69d225fd69db1f6b9e553182ef149092553f246723abf26fdcc5a70b",
+			"digestAlgorithm": "rfc8785-sha256",
+			"executionRef": {
+				"digest": "sha256:42c6816336a75799acbb6db06649ebac5ec2802924f5173f7df692a46f5b9356",
+				"identity": "profile-improvement-fixture-runner",
+				"kind": "agent-profile-improvement-execution-ref"
+			},
+			"kind": "agent-profile-improvement-experiment",
+			"policy": {
+				"bootstrapSeed": 17,
+				"confidenceLevel": .95,
+				"criticalDimensions": [],
+				"deltaThreshold": 0,
+				"minProductiveRuns": 3,
+				"regressionTolerance": 0,
+				"resamples": 100
+			},
+			"source": {
+				"kind": "platform-agent-profile",
+				"sourceDigest": "sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704",
+				"sourceIdentity": "profile-support",
+				"sourceRevision": 7
+			}
+		},
+		"kind": "agent-profile-improvement-measured-comparison",
+		"measurements": [
+			{
+				"baseline": {
+					"billing": [{
+						"digest": "sha256:f8df8abb545f964e51ae5dfbe371db8812ffc67e1ac620467004d1d14622aebb",
+						"identity": "bill-baseline-0",
+						"kind": "platform-billing"
+					}],
+					"digest": "sha256:5f82aa16ab45fb06d00e5f04ecff5f746f2cd31121afb7c2ac31b8cd703ee316",
+					"digestAlgorithm": "rfc8785-sha256",
+					"executionId": "baseline-0",
+					"executionRef": {
+						"digest": "sha256:42c6816336a75799acbb6db06649ebac5ec2802924f5173f7df692a46f5b9356",
+						"identity": "profile-improvement-fixture-runner",
+						"kind": "agent-profile-improvement-execution-ref"
+					},
+					"grading": {
+						"dimensions": [{
+							"name": "quality",
+							"score": .2
+						}],
+						"evidence": {
+							"digest": "sha256:d4df39e172d3701cceb170a3b4f809aebc61c050acc6f0ded51cd178bd8bad79",
+							"identity": "grade-baseline-0",
+							"kind": "agent-eval-grading"
+						},
+						"grader": {
+							"artifact": {
+								"byteLength": 1,
+								"locator": {
+									"bucket": "agent-eval",
+									"key": "graders/profile-quality.json",
+									"kind": "s3",
+									"region": "us-east-1"
+								},
+								"sha256": "sha256:1f576a8a9d7f4af7bf070698b2be73904847d6f28efec0c2366d1f2bc6117e0f"
+							},
+							"format": "tangle-grader",
+							"name": "profile-quality",
+							"version": "1"
+						},
+						"passed": true,
+						"score": .2,
+						"timing": {
+							"durationMs": 10,
+							"endedAtMs": 110,
+							"startedAtMs": 100
+						},
+						"usage": {
+							"cachedInputTokens": 0,
+							"costProvenance": "observed",
+							"costUsdNanos": 10,
+							"inputTokens": 2,
+							"modelCalls": 1,
+							"outputTokens": 1,
+							"reasoningTokens": 0
+						}
+					},
+					"kind": "agent-profile-improvement-run",
+					"limits": {
+						"maxCostUsd": 1,
+						"maxInputTokens": 1e3,
+						"maxModelCalls": 2,
+						"maxOutputTokens": 1e3,
+						"maxSteps": 10,
+						"timeoutMs": 3e4
+					},
+					"outcome": { "status": "succeeded" },
+					"output": {
+						"digest": "sha256:45115d63436a7d83e2f45b791804f24047db2f69377b9796c41b39337110de18",
+						"identity": "output-baseline-0",
+						"kind": "platform-output"
+					},
+					"resolvedModel": {
+						"model": "claude-sonnet-4-6",
+						"provider": "anthropic",
+						"reasoningEffort": "medium",
+						"requested": "anthropic/claude-sonnet-4-6",
+						"snapshot": "2026-06-01"
+					},
+					"runCell": {
+						"arm": "baseline",
+						"attempt": 1,
+						"digest": "sha256:40c8cbe747b60e7a3912a607182340232391b629d41cc6b2c914213f5702088e",
+						"experimentDigest": "sha256:ae081dce69d225fd69db1f6b9e553182ef149092553f246723abf26fdcc5a70b",
+						"kind": "agent-profile-improvement-run-cell",
+						"repetition": 0,
+						"seed": 11,
+						"stateDigest": "sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704",
+						"suiteDigest": "sha256:b625b9c8791b5482f5b4e7c2471eb3ca1b719b5424e86d2bcb1044defbba635e",
+						"taskDigest": "sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e",
+						"taskIndex": 0
+					},
+					"runRecord": {
+						"digest": "sha256:7e8cdb47c8f1c5200bd55f6ecbb05f060d13e93c542978a0cb67eef1f7ea7cd9",
+						"identity": "baseline-0",
+						"kind": "agent-eval-run-record"
+					},
+					"steps": 1,
+					"timing": {
+						"durationMs": 100,
+						"endedAtMs": 100,
+						"startedAtMs": 0
+					},
+					"trace": {
+						"eventCount": 4,
+						"evidence": {
+							"digest": "sha256:7bbb44eebc7f286a293da7309af86cb966ab31a37dc30e9d993523cddfb0a65c",
+							"identity": "trace-baseline-0",
+							"kind": "platform-trace"
+						},
+						"modelCallCount": 1
+					},
+					"usage": {
+						"cachedInputTokens": 0,
+						"costProvenance": "observed",
+						"costUsdNanos": 100,
+						"inputTokens": 10,
+						"modelCalls": 1,
+						"outputTokens": 5,
+						"reasoningTokens": 0
+					}
+				},
+				"candidate": {
+					"billing": [{
+						"digest": "sha256:fc4aebe4eb99e761a95279dcfe7478596a505fd5c5611cfa62d2f1dbe5d98c2f",
+						"identity": "bill-candidate-0",
+						"kind": "platform-billing"
+					}],
+					"digest": "sha256:daf0f586af2ac6a2287e5aefa5b29f7337962168abd5a7c70bb7e6b3837f9b95",
+					"digestAlgorithm": "rfc8785-sha256",
+					"executionId": "candidate-0",
+					"executionRef": {
+						"digest": "sha256:42c6816336a75799acbb6db06649ebac5ec2802924f5173f7df692a46f5b9356",
+						"identity": "profile-improvement-fixture-runner",
+						"kind": "agent-profile-improvement-execution-ref"
+					},
+					"grading": {
+						"dimensions": [{
+							"name": "quality",
+							"score": .6
+						}],
+						"evidence": {
+							"digest": "sha256:d1c4971c7a415aaa1e2a1101611a18aa7c3870411c058cadba35b21fd2faf27a",
+							"identity": "grade-candidate-0",
+							"kind": "agent-eval-grading"
+						},
+						"grader": {
+							"artifact": {
+								"byteLength": 1,
+								"locator": {
+									"bucket": "agent-eval",
+									"key": "graders/profile-quality.json",
+									"kind": "s3",
+									"region": "us-east-1"
+								},
+								"sha256": "sha256:1f576a8a9d7f4af7bf070698b2be73904847d6f28efec0c2366d1f2bc6117e0f"
+							},
+							"format": "tangle-grader",
+							"name": "profile-quality",
+							"version": "1"
+						},
+						"passed": true,
+						"score": .6,
+						"timing": {
+							"durationMs": 10,
+							"endedAtMs": 110,
+							"startedAtMs": 100
+						},
+						"usage": {
+							"cachedInputTokens": 0,
+							"costProvenance": "observed",
+							"costUsdNanos": 10,
+							"inputTokens": 2,
+							"modelCalls": 1,
+							"outputTokens": 1,
+							"reasoningTokens": 0
+						}
+					},
+					"kind": "agent-profile-improvement-run",
+					"limits": {
+						"maxCostUsd": 1,
+						"maxInputTokens": 1e3,
+						"maxModelCalls": 2,
+						"maxOutputTokens": 1e3,
+						"maxSteps": 10,
+						"timeoutMs": 3e4
+					},
+					"outcome": { "status": "succeeded" },
+					"output": {
+						"digest": "sha256:3d57f4f4f14a76c3b7a5a9edc47e8e5d8033896d474a454da74125362456639f",
+						"identity": "output-candidate-0",
+						"kind": "platform-output"
+					},
+					"resolvedModel": {
+						"model": "claude-sonnet-4-6",
+						"provider": "anthropic",
+						"reasoningEffort": "medium",
+						"requested": "anthropic/claude-sonnet-4-6",
+						"snapshot": "2026-06-01"
+					},
+					"runCell": {
+						"arm": "candidate",
+						"attempt": 1,
+						"digest": "sha256:f9fb55d43947f101f10061b11eefd13ae1a56450e515dc6523d07d7261f94221",
+						"experimentDigest": "sha256:ae081dce69d225fd69db1f6b9e553182ef149092553f246723abf26fdcc5a70b",
+						"kind": "agent-profile-improvement-run-cell",
+						"repetition": 0,
+						"seed": 11,
+						"stateDigest": "sha256:103f77bc8481601eef1ad5fe6ba84a40dffabc3a44f421f8c8559121edab84e9",
+						"suiteDigest": "sha256:b625b9c8791b5482f5b4e7c2471eb3ca1b719b5424e86d2bcb1044defbba635e",
+						"taskDigest": "sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e",
+						"taskIndex": 0
+					},
+					"runRecord": {
+						"digest": "sha256:9ea29e94240319216941b87edcc391aed43d45d8081f8b43849591748606e78f",
+						"identity": "candidate-0",
+						"kind": "agent-eval-run-record"
+					},
+					"steps": 1,
+					"timing": {
+						"durationMs": 100,
+						"endedAtMs": 100,
+						"startedAtMs": 0
+					},
+					"trace": {
+						"eventCount": 4,
+						"evidence": {
+							"digest": "sha256:63806d09543cd4337accb01f4806e6825e2fa40b219d0d90ffbcaf369b4cbb5f",
+							"identity": "trace-candidate-0",
+							"kind": "platform-trace"
+						},
+						"modelCallCount": 1
+					},
+					"usage": {
+						"cachedInputTokens": 0,
+						"costProvenance": "observed",
+						"costUsdNanos": 100,
+						"inputTokens": 10,
+						"modelCalls": 1,
+						"outputTokens": 5,
+						"reasoningTokens": 0
+					}
+				}
+			},
+			{
+				"baseline": {
+					"billing": [{
+						"digest": "sha256:0ec2f0978881d5717c3b7bf9ea888b73580693ee5b8e232cb40cb7e57503f263",
+						"identity": "bill-baseline-1",
+						"kind": "platform-billing"
+					}],
+					"digest": "sha256:44788b5fa495aa316b9e730f95dd870f7723171cfec1fff84016126b9346f17b",
+					"digestAlgorithm": "rfc8785-sha256",
+					"executionId": "baseline-1",
+					"executionRef": {
+						"digest": "sha256:42c6816336a75799acbb6db06649ebac5ec2802924f5173f7df692a46f5b9356",
+						"identity": "profile-improvement-fixture-runner",
+						"kind": "agent-profile-improvement-execution-ref"
+					},
+					"grading": {
+						"dimensions": [{
+							"name": "quality",
+							"score": .2
+						}],
+						"evidence": {
+							"digest": "sha256:26079c88be9d35bb6505546daaaae5263a5a86ce7a2958656e95ce77f42d57db",
+							"identity": "grade-baseline-1",
+							"kind": "agent-eval-grading"
+						},
+						"grader": {
+							"artifact": {
+								"byteLength": 1,
+								"locator": {
+									"bucket": "agent-eval",
+									"key": "graders/profile-quality.json",
+									"kind": "s3",
+									"region": "us-east-1"
+								},
+								"sha256": "sha256:1f576a8a9d7f4af7bf070698b2be73904847d6f28efec0c2366d1f2bc6117e0f"
+							},
+							"format": "tangle-grader",
+							"name": "profile-quality",
+							"version": "1"
+						},
+						"passed": true,
+						"score": .2,
+						"timing": {
+							"durationMs": 10,
+							"endedAtMs": 1110,
+							"startedAtMs": 1100
+						},
+						"usage": {
+							"cachedInputTokens": 0,
+							"costProvenance": "observed",
+							"costUsdNanos": 10,
+							"inputTokens": 2,
+							"modelCalls": 1,
+							"outputTokens": 1,
+							"reasoningTokens": 0
+						}
+					},
+					"kind": "agent-profile-improvement-run",
+					"limits": {
+						"maxCostUsd": 1,
+						"maxInputTokens": 1e3,
+						"maxModelCalls": 2,
+						"maxOutputTokens": 1e3,
+						"maxSteps": 10,
+						"timeoutMs": 3e4
+					},
+					"outcome": { "status": "succeeded" },
+					"output": {
+						"digest": "sha256:c3a0a3eab137f6c1e232721d2c7db1237ecb9ec1625447ee1d8d90de0bf2bd33",
+						"identity": "output-baseline-1",
+						"kind": "platform-output"
+					},
+					"resolvedModel": {
+						"model": "claude-sonnet-4-6",
+						"provider": "anthropic",
+						"reasoningEffort": "medium",
+						"requested": "anthropic/claude-sonnet-4-6",
+						"snapshot": "2026-06-01"
+					},
+					"runCell": {
+						"arm": "baseline",
+						"attempt": 1,
+						"digest": "sha256:007bd4c6cadedf64e8792d7eb21a356b2b41b90ed9e04cb0d916bdf20cd2388e",
+						"experimentDigest": "sha256:ae081dce69d225fd69db1f6b9e553182ef149092553f246723abf26fdcc5a70b",
+						"kind": "agent-profile-improvement-run-cell",
+						"repetition": 1,
+						"seed": 12,
+						"stateDigest": "sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704",
+						"suiteDigest": "sha256:b625b9c8791b5482f5b4e7c2471eb3ca1b719b5424e86d2bcb1044defbba635e",
+						"taskDigest": "sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e",
+						"taskIndex": 0
+					},
+					"runRecord": {
+						"digest": "sha256:581e68267991f493474b1f35fc07a95efd9781b55ded5fde4080df55e4ee724b",
+						"identity": "baseline-1",
+						"kind": "agent-eval-run-record"
+					},
+					"steps": 1,
+					"timing": {
+						"durationMs": 100,
+						"endedAtMs": 1100,
+						"startedAtMs": 1e3
+					},
+					"trace": {
+						"eventCount": 4,
+						"evidence": {
+							"digest": "sha256:0be44b67b6482a4d6238a50f98edfd3b406b9a15ca39bc7002fba7fdefaf3a72",
+							"identity": "trace-baseline-1",
+							"kind": "platform-trace"
+						},
+						"modelCallCount": 1
+					},
+					"usage": {
+						"cachedInputTokens": 0,
+						"costProvenance": "observed",
+						"costUsdNanos": 100,
+						"inputTokens": 10,
+						"modelCalls": 1,
+						"outputTokens": 5,
+						"reasoningTokens": 0
+					}
+				},
+				"candidate": {
+					"billing": [{
+						"digest": "sha256:f35d12be88a231cdfaf25ef346a075a2866f93d9fba24dfe6a33948d7d819ef7",
+						"identity": "bill-candidate-1",
+						"kind": "platform-billing"
+					}],
+					"digest": "sha256:688a38e1dae394d7845e892bcee5644297a27e79c1f3813ca29afe30de090365",
+					"digestAlgorithm": "rfc8785-sha256",
+					"executionId": "candidate-1",
+					"executionRef": {
+						"digest": "sha256:42c6816336a75799acbb6db06649ebac5ec2802924f5173f7df692a46f5b9356",
+						"identity": "profile-improvement-fixture-runner",
+						"kind": "agent-profile-improvement-execution-ref"
+					},
+					"grading": {
+						"dimensions": [{
+							"name": "quality",
+							"score": .6
+						}],
+						"evidence": {
+							"digest": "sha256:7ed258fae411110158ff573735ad08e4c0ca32b1e5edf8705a475b8f91aae055",
+							"identity": "grade-candidate-1",
+							"kind": "agent-eval-grading"
+						},
+						"grader": {
+							"artifact": {
+								"byteLength": 1,
+								"locator": {
+									"bucket": "agent-eval",
+									"key": "graders/profile-quality.json",
+									"kind": "s3",
+									"region": "us-east-1"
+								},
+								"sha256": "sha256:1f576a8a9d7f4af7bf070698b2be73904847d6f28efec0c2366d1f2bc6117e0f"
+							},
+							"format": "tangle-grader",
+							"name": "profile-quality",
+							"version": "1"
+						},
+						"passed": true,
+						"score": .6,
+						"timing": {
+							"durationMs": 10,
+							"endedAtMs": 1110,
+							"startedAtMs": 1100
+						},
+						"usage": {
+							"cachedInputTokens": 0,
+							"costProvenance": "observed",
+							"costUsdNanos": 10,
+							"inputTokens": 2,
+							"modelCalls": 1,
+							"outputTokens": 1,
+							"reasoningTokens": 0
+						}
+					},
+					"kind": "agent-profile-improvement-run",
+					"limits": {
+						"maxCostUsd": 1,
+						"maxInputTokens": 1e3,
+						"maxModelCalls": 2,
+						"maxOutputTokens": 1e3,
+						"maxSteps": 10,
+						"timeoutMs": 3e4
+					},
+					"outcome": { "status": "succeeded" },
+					"output": {
+						"digest": "sha256:abe4d04fa6375d75469aa41839eb34758c4a5cdf17b204a2294b1d1a9faa8ffb",
+						"identity": "output-candidate-1",
+						"kind": "platform-output"
+					},
+					"resolvedModel": {
+						"model": "claude-sonnet-4-6",
+						"provider": "anthropic",
+						"reasoningEffort": "medium",
+						"requested": "anthropic/claude-sonnet-4-6",
+						"snapshot": "2026-06-01"
+					},
+					"runCell": {
+						"arm": "candidate",
+						"attempt": 1,
+						"digest": "sha256:abad51b565dd3e740b87d260a553318eeb1b5907938841a5942b0df629aadfc3",
+						"experimentDigest": "sha256:ae081dce69d225fd69db1f6b9e553182ef149092553f246723abf26fdcc5a70b",
+						"kind": "agent-profile-improvement-run-cell",
+						"repetition": 1,
+						"seed": 12,
+						"stateDigest": "sha256:103f77bc8481601eef1ad5fe6ba84a40dffabc3a44f421f8c8559121edab84e9",
+						"suiteDigest": "sha256:b625b9c8791b5482f5b4e7c2471eb3ca1b719b5424e86d2bcb1044defbba635e",
+						"taskDigest": "sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e",
+						"taskIndex": 0
+					},
+					"runRecord": {
+						"digest": "sha256:dafe5fbae73dab4b49b3411fc26cc436ddc463dfd0ef86ee23c8b0a91c81be67",
+						"identity": "candidate-1",
+						"kind": "agent-eval-run-record"
+					},
+					"steps": 1,
+					"timing": {
+						"durationMs": 100,
+						"endedAtMs": 1100,
+						"startedAtMs": 1e3
+					},
+					"trace": {
+						"eventCount": 4,
+						"evidence": {
+							"digest": "sha256:6a33d6b9699ebf78e92845bdae506a3b37e9598a3ead4b5f01676d9573c2fc1a",
+							"identity": "trace-candidate-1",
+							"kind": "platform-trace"
+						},
+						"modelCallCount": 1
+					},
+					"usage": {
+						"cachedInputTokens": 0,
+						"costProvenance": "observed",
+						"costUsdNanos": 100,
+						"inputTokens": 10,
+						"modelCalls": 1,
+						"outputTokens": 5,
+						"reasoningTokens": 0
+					}
+				}
+			},
+			{
+				"baseline": {
+					"billing": [{
+						"digest": "sha256:6365caecc9b3f4a7d7fa3b3a7efea8a76fc50d93129b02f7ba448ceee6b806a1",
+						"identity": "bill-baseline-2",
+						"kind": "platform-billing"
+					}],
+					"digest": "sha256:07f982ab59585df9b30fe498ce3969134625d5d3a5cc4cb3549cca83175ef2d4",
+					"digestAlgorithm": "rfc8785-sha256",
+					"executionId": "baseline-2",
+					"executionRef": {
+						"digest": "sha256:42c6816336a75799acbb6db06649ebac5ec2802924f5173f7df692a46f5b9356",
+						"identity": "profile-improvement-fixture-runner",
+						"kind": "agent-profile-improvement-execution-ref"
+					},
+					"grading": {
+						"dimensions": [{
+							"name": "quality",
+							"score": .2
+						}],
+						"evidence": {
+							"digest": "sha256:2e6344b8ab46798df75c6c8bd2e9c1c4ba5371cfdb84ef12916226a06ee0bffc",
+							"identity": "grade-baseline-2",
+							"kind": "agent-eval-grading"
+						},
+						"grader": {
+							"artifact": {
+								"byteLength": 1,
+								"locator": {
+									"bucket": "agent-eval",
+									"key": "graders/profile-quality.json",
+									"kind": "s3",
+									"region": "us-east-1"
+								},
+								"sha256": "sha256:1f576a8a9d7f4af7bf070698b2be73904847d6f28efec0c2366d1f2bc6117e0f"
+							},
+							"format": "tangle-grader",
+							"name": "profile-quality",
+							"version": "1"
+						},
+						"passed": true,
+						"score": .2,
+						"timing": {
+							"durationMs": 10,
+							"endedAtMs": 2110,
+							"startedAtMs": 2100
+						},
+						"usage": {
+							"cachedInputTokens": 0,
+							"costProvenance": "observed",
+							"costUsdNanos": 10,
+							"inputTokens": 2,
+							"modelCalls": 1,
+							"outputTokens": 1,
+							"reasoningTokens": 0
+						}
+					},
+					"kind": "agent-profile-improvement-run",
+					"limits": {
+						"maxCostUsd": 1,
+						"maxInputTokens": 1e3,
+						"maxModelCalls": 2,
+						"maxOutputTokens": 1e3,
+						"maxSteps": 10,
+						"timeoutMs": 3e4
+					},
+					"outcome": { "status": "succeeded" },
+					"output": {
+						"digest": "sha256:7ecb1cb5e93f0154d39393745ca59c763725073a528396dacc06703ff6559a72",
+						"identity": "output-baseline-2",
+						"kind": "platform-output"
+					},
+					"resolvedModel": {
+						"model": "claude-sonnet-4-6",
+						"provider": "anthropic",
+						"reasoningEffort": "medium",
+						"requested": "anthropic/claude-sonnet-4-6",
+						"snapshot": "2026-06-01"
+					},
+					"runCell": {
+						"arm": "baseline",
+						"attempt": 1,
+						"digest": "sha256:b4f83f1ba81c01032326dba7523e68a41c50cdc2447452aff480b75dfa1e8894",
+						"experimentDigest": "sha256:ae081dce69d225fd69db1f6b9e553182ef149092553f246723abf26fdcc5a70b",
+						"kind": "agent-profile-improvement-run-cell",
+						"repetition": 2,
+						"seed": 13,
+						"stateDigest": "sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704",
+						"suiteDigest": "sha256:b625b9c8791b5482f5b4e7c2471eb3ca1b719b5424e86d2bcb1044defbba635e",
+						"taskDigest": "sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e",
+						"taskIndex": 0
+					},
+					"runRecord": {
+						"digest": "sha256:136ff81d5b2a688db3fbe00bbfabb593e6179d9e71142d687ef38208b2301cd3",
+						"identity": "baseline-2",
+						"kind": "agent-eval-run-record"
+					},
+					"steps": 1,
+					"timing": {
+						"durationMs": 100,
+						"endedAtMs": 2100,
+						"startedAtMs": 2e3
+					},
+					"trace": {
+						"eventCount": 4,
+						"evidence": {
+							"digest": "sha256:a9c52e32611334aa710c0930b36aa3a2faf45dfdb77f083e9e4c02ab67b2812e",
+							"identity": "trace-baseline-2",
+							"kind": "platform-trace"
+						},
+						"modelCallCount": 1
+					},
+					"usage": {
+						"cachedInputTokens": 0,
+						"costProvenance": "observed",
+						"costUsdNanos": 100,
+						"inputTokens": 10,
+						"modelCalls": 1,
+						"outputTokens": 5,
+						"reasoningTokens": 0
+					}
+				},
+				"candidate": {
+					"billing": [{
+						"digest": "sha256:f5f6023645c3d531f29eb0b331787117a1c87d1cd5d53f434ab4a5ab8eb1caba",
+						"identity": "bill-candidate-2",
+						"kind": "platform-billing"
+					}],
+					"digest": "sha256:4daa28e364e711b0896e7f8bbfc9153aaa4e997672578ac0a1cd6dcdeb2148e0",
+					"digestAlgorithm": "rfc8785-sha256",
+					"executionId": "candidate-2",
+					"executionRef": {
+						"digest": "sha256:42c6816336a75799acbb6db06649ebac5ec2802924f5173f7df692a46f5b9356",
+						"identity": "profile-improvement-fixture-runner",
+						"kind": "agent-profile-improvement-execution-ref"
+					},
+					"grading": {
+						"dimensions": [{
+							"name": "quality",
+							"score": .6
+						}],
+						"evidence": {
+							"digest": "sha256:9046047f8a5ad8cda1e059047b3bac42abb438a8d9f9a60c0a139af5375dc313",
+							"identity": "grade-candidate-2",
+							"kind": "agent-eval-grading"
+						},
+						"grader": {
+							"artifact": {
+								"byteLength": 1,
+								"locator": {
+									"bucket": "agent-eval",
+									"key": "graders/profile-quality.json",
+									"kind": "s3",
+									"region": "us-east-1"
+								},
+								"sha256": "sha256:1f576a8a9d7f4af7bf070698b2be73904847d6f28efec0c2366d1f2bc6117e0f"
+							},
+							"format": "tangle-grader",
+							"name": "profile-quality",
+							"version": "1"
+						},
+						"passed": true,
+						"score": .6,
+						"timing": {
+							"durationMs": 10,
+							"endedAtMs": 2110,
+							"startedAtMs": 2100
+						},
+						"usage": {
+							"cachedInputTokens": 0,
+							"costProvenance": "observed",
+							"costUsdNanos": 10,
+							"inputTokens": 2,
+							"modelCalls": 1,
+							"outputTokens": 1,
+							"reasoningTokens": 0
+						}
+					},
+					"kind": "agent-profile-improvement-run",
+					"limits": {
+						"maxCostUsd": 1,
+						"maxInputTokens": 1e3,
+						"maxModelCalls": 2,
+						"maxOutputTokens": 1e3,
+						"maxSteps": 10,
+						"timeoutMs": 3e4
+					},
+					"outcome": { "status": "succeeded" },
+					"output": {
+						"digest": "sha256:58439d73500b422267dfc9d873a9132aa384ebbf706b75fdfa69793b70f65d65",
+						"identity": "output-candidate-2",
+						"kind": "platform-output"
+					},
+					"resolvedModel": {
+						"model": "claude-sonnet-4-6",
+						"provider": "anthropic",
+						"reasoningEffort": "medium",
+						"requested": "anthropic/claude-sonnet-4-6",
+						"snapshot": "2026-06-01"
+					},
+					"runCell": {
+						"arm": "candidate",
+						"attempt": 1,
+						"digest": "sha256:5f66650dc0f5238aeaca52750577fc2b280b8769e0b4b28d592a9a8173bd712a",
+						"experimentDigest": "sha256:ae081dce69d225fd69db1f6b9e553182ef149092553f246723abf26fdcc5a70b",
+						"kind": "agent-profile-improvement-run-cell",
+						"repetition": 2,
+						"seed": 13,
+						"stateDigest": "sha256:103f77bc8481601eef1ad5fe6ba84a40dffabc3a44f421f8c8559121edab84e9",
+						"suiteDigest": "sha256:b625b9c8791b5482f5b4e7c2471eb3ca1b719b5424e86d2bcb1044defbba635e",
+						"taskDigest": "sha256:7e5db980d166e67bc366f2f08e25cab5a2f8558530744f6755ee5997ef345c1e",
+						"taskIndex": 0
+					},
+					"runRecord": {
+						"digest": "sha256:e3c9377d8fc6c3f427f8ba79a2e2cfec2319d2cf2e923f1344db26195780bc0f",
+						"identity": "candidate-2",
+						"kind": "agent-eval-run-record"
+					},
+					"steps": 1,
+					"timing": {
+						"durationMs": 100,
+						"endedAtMs": 2100,
+						"startedAtMs": 2e3
+					},
+					"trace": {
+						"eventCount": 4,
+						"evidence": {
+							"digest": "sha256:7b862fccf82622a8c3a5b0816f5a543ba61454972e8799219948df150fd0a7fa",
+							"identity": "trace-candidate-2",
+							"kind": "platform-trace"
+						},
+						"modelCallCount": 1
+					},
+					"usage": {
+						"cachedInputTokens": 0,
+						"costProvenance": "observed",
+						"costUsdNanos": 100,
+						"inputTokens": 10,
+						"modelCalls": 1,
+						"outputTokens": 5,
+						"reasoningTokens": 0
+					}
+				}
+			}
+		],
+		"metadata": {
+			"fixture": "agent-profile-improvement-proposal",
+			"runtimeVersion": "0.107.3"
+		},
+		"objectives": [
+			{
+				"availability": "measured",
+				"baseline": .2,
+				"candidate": .6,
+				"confidenceInterval": {
+					"level": .95,
+					"lower": .29999999999999993,
+					"method": "paired-bootstrap",
+					"resamples": 100,
+					"statistic": "mean",
+					"upper": .5
+				},
+				"delta": .39999999999999997,
+				"direction": "higher-is-better",
+				"kind": "objective",
+				"n": 3,
+				"name": "quality",
+				"unit": "score"
+			},
+			{
+				"availability": "measured",
+				"baseline": .2,
+				"candidate": .6,
+				"confidenceInterval": {
+					"level": .95,
+					"lower": .29999999999999993,
+					"method": "paired-bootstrap",
+					"resamples": 100,
+					"statistic": "mean",
+					"upper": .5
+				},
+				"delta": .39999999999999997,
+				"direction": "higher-is-better",
+				"kind": "dimension",
+				"n": 3,
+				"name": "quality",
+				"objective": "quality",
+				"unit": "score"
+			},
+			{
+				"availability": "measured",
+				"baseline": 11e-8,
+				"candidate": 11e-8,
+				"confidenceInterval": {
+					"level": .95,
+					"lower": -.1,
+					"method": "paired-bootstrap",
+					"resamples": 100,
+					"statistic": "mean",
+					"upper": .1
+				},
+				"delta": 0,
+				"direction": "lower-is-better",
+				"kind": "cost",
+				"n": 3,
+				"name": "cost",
+				"unit": "usd"
+			},
+			{
+				"availability": "measured",
+				"baseline": 110,
+				"candidate": 110,
+				"confidenceInterval": {
+					"level": .95,
+					"lower": -.1,
+					"method": "paired-bootstrap",
+					"resamples": 100,
+					"statistic": "mean",
+					"upper": .1
+				},
+				"delta": 0,
+				"direction": "lower-is-better",
+				"kind": "latency",
+				"n": 3,
+				"name": "latency",
+				"unit": "milliseconds"
+			}
+		],
+		"overall": {
+			"baseline": .2,
+			"candidate": .6,
+			"confidenceInterval": {
+				"level": .95,
+				"lower": .29999999999999993,
+				"method": "paired-bootstrap",
+				"resamples": 100,
+				"statistic": "mean",
+				"upper": .5
+			},
+			"delta": .39999999999999997,
+			"direction": "higher-is-better",
+			"n": 3,
+			"name": "composite",
+			"unit": "score"
+		},
+		"power": {
+			"confidenceLevel": .95,
+			"minimumDetectableDelta": .1,
+			"n": 3,
+			"reason": "three paired held-out runs",
+			"scaleAssumed": true,
+			"sharedScorerChannel": true,
+			"sufficient": true
+		},
+		"provenance": {
+			"baselineContentHash": "sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704",
+			"candidateContentHash": "sha256:103f77bc8481601eef1ad5fe6ba84a40dffabc3a44f421f8c8559121edab84e9",
+			"kind": "agent-eval-loop",
+			"recordDigest": "sha256:44465d9d1b5b2f7697ad7a4617d7da80ccad97538b55e8a41b016cddc9c66bcf",
+			"runId": "profile-improvement-1",
+			"schema": "agent-eval/profile-matrix/v1"
+		}
+	},
+	findings: [],
+	kind: "agent-improvement-proposal",
+	proposedAt: "2026-07-10T01:30:00.000Z",
+	runId: "profile-improvement-1"
+};
+//#endregion
+//#region src/testing/fixtures/agent-profile-improvement-state.json
+var agent_profile_improvement_state_default = {
+	baselineProfile: {
+		"name": "support-agent",
+		"prompt": { "systemPrompt": "Answer directly." },
+		"tools": { "Read": true },
+		"resources": { "skills": [{
+			"kind": "inline",
+			"name": "support.SKILL.md",
+			"content": "Use the support case context."
+		}] }
+	},
+	candidateProfile: {
+		"name": "support-agent",
+		"prompt": { "systemPrompt": "Answer directly, cite the source, and state uncertainty." },
+		"tools": { "Read": true },
+		"resources": { "skills": [{
+			"kind": "inline",
+			"name": "support.SKILL.md",
+			"content": "Use the support case context."
+		}, {
+			"kind": "inline",
+			"name": "sources.SKILL.md",
+			"content": "Cite the evidence you use."
+		}] }
+	},
+	recommendedSize: "small"
+};
+//#endregion
+//#region src/testing/index.ts
+const serializedAgentImprovementProposalFixture = JSON.stringify(agent_improvement_proposal_default);
+const serializedAgentProfileImprovementProposalFixture = JSON.stringify(agent_profile_improvement_proposal_default);
+const serializedAgentProfileImprovementStateFixture = JSON.stringify(agent_profile_improvement_state_default);
 /** Load an isolated, production-validated Runtime proposal for consumer tests. */
 function loadAgentImprovementProposalFixture() {
 	return verifyAgentImprovementProposal(JSON.parse(serializedAgentImprovementProposalFixture));
 }
+/** Load an isolated profile proposal and its private activation state for consumer tests. */
+function loadAgentProfileImprovementFixture() {
+	const proposal = loadAgentProfileImprovementProposal();
+	const state = parseAgentProfileImprovementState(JSON.parse(serializedAgentProfileImprovementStateFixture));
+	const experiment = proposal.evaluation.experiment;
+	const baselineStateDigest = profileStateDigest(state.baselineProfile, state.recommendedSize);
+	const candidateStateDigest = profileStateDigest(state.candidateProfile, state.recommendedSize);
+	if (baselineStateDigest !== experiment.baseline.stateDigest || candidateStateDigest !== experiment.candidate.stateDigest) throw new Error("profile improvement fixture state does not match its proposal");
+	if (canonicalCandidateDigest$1(experiment.change.reduce((profile, change, index) => applyExactAgentProfileDiff(profile, change, `profile improvement fixture change ${index}`), state.baselineProfile)) !== canonicalCandidateDigest$1(state.candidateProfile)) throw new Error("profile improvement fixture candidate does not match its proposal changes");
+	return immutableCandidateValue({
+		proposal,
+		...state
+	});
+}
+function loadAgentProfileImprovementProposal() {
+	const proposal = verifyAgentImprovementProposal(JSON.parse(serializedAgentProfileImprovementProposalFixture));
+	if (proposal.evaluation.kind !== "agent-profile-improvement-measured-comparison") throw new Error("profile improvement fixture contains a sealed candidate comparison");
+	return {
+		...proposal,
+		evaluation: proposal.evaluation
+	};
+}
+function parseAgentProfileImprovementState(input) {
+	if (typeof input !== "object" || input === null || Array.isArray(input)) throw new Error("profile improvement fixture state must be an object");
+	const record = input;
+	const fields = Object.keys(record).sort();
+	const expectedFields = [
+		"baselineProfile",
+		"candidateProfile",
+		"recommendedSize"
+	];
+	if (fields.length !== expectedFields.length || fields.some((field, i) => field !== expectedFields[i])) throw new Error("profile improvement fixture state contains unsupported fields");
+	const recommendedSize = SANDBOX_SIZE_PRESET_NAMES.find((size) => size === record.recommendedSize);
+	if (!recommendedSize) throw new Error("profile improvement fixture has an invalid recommended size");
+	return {
+		baselineProfile: parseExactAgentProfile(record.baselineProfile, "profile improvement fixture baseline"),
+		candidateProfile: parseExactAgentProfile(record.candidateProfile, "profile improvement fixture candidate"),
+		recommendedSize
+	};
+}
+function profileStateDigest(profile, recommendedSize) {
+	return canonicalCandidateDigest$1({
+		definition: profile,
+		recommendedSize
+	});
+}
 //#endregion
-export { loadAgentImprovementProposalFixture };
+export { loadAgentImprovementProposalFixture, loadAgentProfileImprovementFixture };
 //# sourceMappingURL=testing.js.map

package/dist/testing.js.map CHANGED Viewed

	@@ -1 +1 @@
1	- {"version":3,"file":"testing.js","names":["canonicalAgentImprovementProposalFixture"],"sources":["../src/testing/fixtures/agent-improvement-proposal.json","../src/testing/index.ts"],"sourcesContent":["","import type { AgentImprovementProposal } from '@tangle-network/agent-interface'\n\nimport { verifyAgentImprovementProposal } from '../intelligence/improvement-cycle'\nimport canonicalAgentImprovementProposalFixture from './fixtures/agent-improvement-proposal.json'\n\nconst serializedAgentImprovementProposalFixture = JSON.stringify(\n canonicalAgentImprovementProposalFixture,\n)\n\n/** Load an isolated, production-validated Runtime proposal for consumer tests. */\nexport function loadAgentImprovementProposalFixture(): AgentImprovementProposal {\n return verifyAgentImprovementProposal(JSON.parse(serializedAgentImprovementProposalFixture))\n}\n"],"mappings":";;;ACKA,MAAM,4CAA4C,KAAK,UACrDA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAAA,CACF;;AAGA,SAAgB,sCAAgE;CAC9E,OAAO,+BAA+B,KAAK,MAAM,yCAAyC,CAAC;AAC7F"}
1	+ {"version":3,"file":"testing.js","names":["canonicalAgentImprovementProposalFixture","canonicalAgentProfileImprovementProposalFixture","canonicalAgentProfileImprovementStateFixture","canonicalCandidateDigest"],"sources":["../src/testing/fixtures/agent-improvement-proposal.json","../src/testing/fixtures/agent-profile-improvement-proposal.json","../src/testing/fixtures/agent-profile-improvement-state.json","../src/testing/index.ts"],"sourcesContent":["","","","import {\n type AgentImprovementProposal,\n type AgentProfile,\n type AgentProfileImprovementMeasuredComparison,\n SANDBOX_SIZE_PRESET_NAMES,\n type SandboxSizePreset,\n} from '@tangle-network/agent-interface'\n\nimport { canonicalCandidateDigest, immutableCandidateValue } from '../candidate-execution/digest'\nimport { applyExactAgentProfileDiff, parseExactAgentProfile } from '../candidate-execution/profile'\nimport { verifyAgentImprovementProposal } from '../intelligence/improvement-cycle'\nimport canonicalAgentImprovementProposalFixture from './fixtures/agent-improvement-proposal.json'\nimport canonicalAgentProfileImprovementProposalFixture from './fixtures/agent-profile-improvement-proposal.json'\nimport canonicalAgentProfileImprovementStateFixture from './fixtures/agent-profile-improvement-state.json'\n\nconst serializedAgentImprovementProposalFixture = JSON.stringify(\n canonicalAgentImprovementProposalFixture,\n)\nconst serializedAgentProfileImprovementProposalFixture = JSON.stringify(\n canonicalAgentProfileImprovementProposalFixture,\n)\nconst serializedAgentProfileImprovementStateFixture = JSON.stringify(\n canonicalAgentProfileImprovementStateFixture,\n)\n\n/** A proposal produced by Runtime's opaque profile-improvement path. /\nexport type AgentProfileImprovementProposalFixture = Omit<\n AgentImprovementProposal,\n 'evaluation'\n> & {\n evaluation: AgentProfileImprovementMeasuredComparison\n}\n\n/* Complete private state for exercising profile activation and restore in consumer tests. /\nexport interface AgentProfileImprovementFixture {\n proposal: AgentProfileImprovementProposalFixture\n baselineProfile: AgentProfile\n candidateProfile: AgentProfile\n recommendedSize: SandboxSizePreset\n}\n\n/* Load an isolated, production-validated Runtime proposal for consumer tests. /\nexport function loadAgentImprovementProposalFixture(): AgentImprovementProposal {\n return verifyAgentImprovementProposal(JSON.parse(serializedAgentImprovementProposalFixture))\n}\n\n/* Load an isolated profile proposal and its private activation state for consumer tests. */\nexport function loadAgentProfileImprovementFixture(): AgentProfileImprovementFixture {\n const proposal = loadAgentProfileImprovementProposal()\n const state = parseAgentProfileImprovementState(\n JSON.parse(serializedAgentProfileImprovementStateFixture),\n )\n const experiment = proposal.evaluation.experiment\n const baselineStateDigest = profileStateDigest(state.baselineProfile, state.recommendedSize)\n const candidateStateDigest = profileStateDigest(state.candidateProfile, state.recommendedSize)\n if (\n baselineStateDigest !== experiment.baseline.stateDigest \|\|\n candidateStateDigest !== experiment.candidate.stateDigest\n ) {\n throw new Error('profile improvement fixture state does not match its proposal')\n }\n const appliedCandidate = experiment.change.reduce(\n (profile, change, index) =>\n applyExactAgentProfileDiff(profile, change, `profile improvement fixture change ${index}`),\n state.baselineProfile,\n )\n if (\n canonicalCandidateDigest(appliedCandidate) !== canonicalCandidateDigest(state.candidateProfile)\n ) {\n throw new Error('profile improvement fixture candidate does not match its proposal changes')\n }\n return immutableCandidateValue({ proposal, ...state })\n}\n\nfunction loadAgentProfileImprovementProposal(): AgentProfileImprovementProposalFixture {\n const proposal = verifyAgentImprovementProposal(\n JSON.parse(serializedAgentProfileImprovementProposalFixture),\n )\n if (proposal.evaluation.kind !== 'agent-profile-improvement-measured-comparison') {\n throw new Error('profile improvement fixture contains a sealed candidate comparison')\n }\n return { ...proposal, evaluation: proposal.evaluation }\n}\n\nfunction parseAgentProfileImprovementState(\n input: unknown,\n): Omit<AgentProfileImprovementFixture, 'proposal'> {\n if (typeof input !== 'object' \|\| input === null \|\| Array.isArray(input)) {\n throw new Error('profile improvement fixture state must be an object')\n }\n const record = input as Record<string, unknown>\n const fields = Object.keys(record).sort()\n const expectedFields = ['baselineProfile', 'candidateProfile', 'recommendedSize']\n if (\n fields.length !== expectedFields.length \|\|\n fields.some((field, i) => field !== expectedFields[i])\n ) {\n throw new Error('profile improvement fixture state contains unsupported fields')\n }\n const recommendedSize = SANDBOX_SIZE_PRESET_NAMES.find((size) => size === record.recommendedSize)\n if (!recommendedSize) {\n throw new Error('profile improvement fixture has an invalid recommended size')\n }\n return {\n baselineProfile: parseExactAgentProfile(\n record.baselineProfile,\n 'profile improvement fixture baseline',\n ),\n candidateProfile: parseExactAgentProfile(\n record.candidateProfile,\n 'profile improvement fixture candidate',\n ),\n recommendedSize,\n }\n}\n\nfunction profileStateDigest(\n profile: AgentProfile,\n recommendedSize: SandboxSizePreset,\n): ReturnType<typeof canonicalCandidateDigest> {\n return canonicalCandidateDigest({ definition: profile, recommendedSize })\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AGeA,MAAM,4CAA4C,KAAK,UACrDA,kCACF;AACA,MAAM,mDAAmD,KAAK,UAC5DC,0CACF;AACA,MAAM,gDAAgD,KAAK,UACzDC,uCACF;;AAmBA,SAAgB,sCAAgE;CAC9E,OAAO,+BAA+B,KAAK,MAAM,yCAAyC,CAAC;AAC7F;;AAGA,SAAgB,qCAAqE;CACnF,MAAM,WAAW,oCAAoC;CACrD,MAAM,QAAQ,kCACZ,KAAK,MAAM,6CAA6C,CAC1D;CACA,MAAM,aAAa,SAAS,WAAW;CACvC,MAAM,sBAAsB,mBAAmB,MAAM,iBAAiB,MAAM,eAAe;CAC3F,MAAM,uBAAuB,mBAAmB,MAAM,kBAAkB,MAAM,eAAe;CAC7F,IACE,wBAAwB,WAAW,SAAS,eAC5C,yBAAyB,WAAW,UAAU,aAE9C,MAAM,IAAI,MAAM,+DAA+D;CAOjF,IACEC,2BANuB,WAAW,OAAO,QACxC,SAAS,QAAQ,UAChB,2BAA2B,SAAS,QAAQ,sCAAsC,OAAO,GAC3F,MAAM,eAGkC,CAAC,MAAMA,2BAAyB,MAAM,gBAAgB,GAE9F,MAAM,IAAI,MAAM,2EAA2E;CAE7F,OAAO,wBAAwB;EAAE;EAAU,GAAG;CAAM,CAAC;AACvD;AAEA,SAAS,sCAA8E;CACrF,MAAM,WAAW,+BACf,KAAK,MAAM,gDAAgD,CAC7D;CACA,IAAI,SAAS,WAAW,SAAS,iDAC/B,MAAM,IAAI,MAAM,oEAAoE;CAEtF,OAAO;EAAE,GAAG;EAAU,YAAY,SAAS;CAAW;AACxD;AAEA,SAAS,kCACP,OACkD;CAClD,IAAI,OAAO,UAAU,YAAY,UAAU,QAAQ,MAAM,QAAQ,KAAK,GACpE,MAAM,IAAI,MAAM,qDAAqD;CAEvE,MAAM,SAAS;CACf,MAAM,SAAS,OAAO,KAAK,MAAM,CAAC,CAAC,KAAK;CACxC,MAAM,iBAAiB;EAAC;EAAmB;EAAoB;CAAiB;CAChF,IACE,OAAO,WAAW,eAAe,UACjC,OAAO,MAAM,OAAO,MAAM,UAAU,eAAe,EAAE,GAErD,MAAM,IAAI,MAAM,+DAA+D;CAEjF,MAAM,kBAAkB,0BAA0B,MAAM,SAAS,SAAS,OAAO,eAAe;CAChG,IAAI,CAAC,iBACH,MAAM,IAAI,MAAM,6DAA6D;CAE/E,OAAO;EACL,iBAAiB,uBACf,OAAO,iBACP,sCACF;EACA,kBAAkB,uBAChB,OAAO,kBACP,uCACF;EACA;CACF;AACF;AAEA,SAAS,mBACP,SACA,iBAC6C;CAC7C,OAAOA,2BAAyB;EAAE,YAAY;EAAS;CAAgB,CAAC;AAC1E"}

package/package.json CHANGED Viewed

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-runtime",
-  "version": "0.107.2",
+  "version": "0.107.3",
   "description": "Shared task-lifecycle skeleton for agents: a recursive loop kernel for chat turns, one-shot tasks, and multi-attempt loops, with trace capture and eval-gated self-improvement. Domain behavior lives in adapters; scoring and ship-gates in @tangle-network/agent-eval.",
   "homepage": "https://github.com/tangle-network/agent-runtime#readme",
   "repository": {
@@ -101,9 +101,9 @@
   "devDependencies": {
     "@arethetypeswrong/cli": "0.18.5",
     "@biomejs/biome": "^2.5.5",
-    "@tangle-network/agent-eval": "0.133.0",
+    "@tangle-network/agent-eval": "0.133.1",
     "@tangle-network/agent-interface": "0.36.0",
-    "@tangle-network/sandbox": "0.15.0",
+    "@tangle-network/sandbox": "0.15.1",
     "@types/node": "26.1.1",
     "@types/tar-stream": "3.1.4",
     "playwright": "^1.62.0",
@@ -120,7 +120,7 @@
   },
   "license": "MIT",
   "peerDependencies": {
-    "@tangle-network/agent-eval": ">=0.133.0 <0.134.0",
+    "@tangle-network/agent-eval": ">=0.133.1 <0.134.0",
     "@tangle-network/agent-interface": ">=0.36.0 <0.37.0",
     "@tangle-network/sandbox": ">=0.15.0 <0.16.0",
     "playwright": "^1.40.0"
@@ -134,7 +134,7 @@
     }
   },
   "dependencies": {
-    "@tangle-network/agent-knowledge": "6.1.2",
+    "@tangle-network/agent-knowledge": "6.1.3",
     "@tangle-network/agent-profile-materialize": "0.9.0",
     "tar-stream": "3.2.0"
   },
@@ -147,12 +147,13 @@
     "lint:fix": "biome check --write src tests examples",
     "typecheck": "tsc --noEmit && pnpm run typecheck:examples",
     "typecheck:examples": "tsc --noEmit -p tsconfig.examples.json",
-    "generate:testing-fixture": "tsx scripts/generate-agent-improvement-proposal-fixture.ts",
-    "check:testing-fixture": "tsx scripts/generate-agent-improvement-proposal-fixture.ts --check",
+    "generate:testing-fixture": "tsx scripts/generate-agent-improvement-proposal-fixtures.ts",
+    "check:testing-fixture": "tsx scripts/generate-agent-improvement-proposal-fixtures.ts --check",
     "check:skills": "node scripts/check-skills.mjs",
     "release:prepare": "node scripts/prepare-release.mjs",
     "verify:bench": "pnpm build && pnpm --filter @tangle-network/agent-bench run typecheck:public && pnpm --filter @tangle-network/agent-bench test && pnpm --filter @tangle-network/agent-bench run verify:package:local-runtime",
     "verify:bench:published": "pnpm build && pnpm --filter @tangle-network/agent-bench run typecheck:public && pnpm --filter @tangle-network/agent-bench test && pnpm --filter @tangle-network/agent-bench run verify:package",
+    "verify:cohort": "node scripts/verify-packed-cohort.mjs",
     "verify:package": "pnpm run build && pnpm run check:testing-fixture && pnpm run check:skills && publint && attw --pack --profile esm-only . && node scripts/verify-package-exports.mjs",
     "verify:official-optimizers": "node scripts/verify-official-optimizers.mjs",
     "verify:primeintellect": "pnpm build && node scripts/verify-primeintellect.mjs",