{ "adoptionRecords": [], "asOf": "2026-08-20T19:08:46Z", "boundaries": { "adoptionRecorded": false, "artifactAcquisitionPerformed": false, "artifactExecutionPerformed": false, "automaticAction": false, "buildNetworkUsed": false, "credentialAccessed": false, "evaluationResultRecorded": false, "fullTextIncluded": false, "guestAdmitted": false, "metadataTitlesAreScientificClaims": false, "moduleCreated": false, "preflightPerformed": false, "scalarScoreUsed": false, "signatureCreated": false, "trialAuthorityIssued": false, "trialRunRecorded": false, "useAuthorityIssued": false }, "contractVersion": "0.1.0", "counts": { "adoptionRecords": 0, "evaluationPlans": 3, "evaluationRequests": 3, "evaluationResults": 0, "preflights": 0, "referrals": 4, "trialAuthorities": 0, "trialRuns": 0, "useAuthorities": 0 }, "evaluationPlans": [ { "authorship": { "compiledBy": "codex-assisted-publication", "exactTextHumanReviewed": false, "humanDirectionRecorded": true, "origin": "maintainer-authorized-assisted-draft" }, "controls": [ "Freeze one self-authored registry, planted-change labels, age/activity/category strata, audit budget, and all policies before a future comparison.", "Give uniform, recency-only, score-only, and coverage-balanced policies the identical audit budget.", "Keep one separately authored fixture held out until policies and measures are frozen." ], "criterionRules": [ { "condition": "Coverage-balanced review exposes more planted change patterns than every baseline under the same budget on the held-out fixture.", "interpretation": "retain-as-bounded-comparative-evidence" }, { "condition": "One or more baselines expose an equal or greater number of planted patterns.", "interpretation": "retain-as-no-observed-increment" }, { "condition": "Budgets differ, labels are ambiguous, a policy is tuned after held-out inspection, or a stratum has no auditable representation.", "interpretation": "retain-as-design-conflict" } ], "draftedOn": "2026-08-20", "evaluationLens": "measurement-reproducibility", "evaluationQuestion": "Under one fixed audit budget, can a coverage-aware re-audit policy expose more planted change patterns across a finite synthetic registry than recency-only, score-only, or uniform baselines?", "expiresAt": "2026-09-20T00:00:00Z", "expiryEffect": "FREEZE_AND_REQUIRE_NEW_HUMAN_REVIEW", "fixtures": [ { "description": "A frozen finite registry with planted changes distributed across age, activity, and category strata for policy development.", "heldOut": false, "id": "synthetic-registry-development", "synthetic": true }, { "description": "A separately authored finite registry with undisclosed planted labels until policies, budgets, and measures are frozen.", "heldOut": true, "id": "synthetic-registry-held-out", "synthetic": true } ], "id": "evaluation-plan:mcp-registry-drift:2026-08-20", "inputClass": "SELF-AUTHORED-SYNTHETIC-ONLY", "measures": [ { "name": "planted-change-patterns-exposed", "rule": "Report the distinct pre-registered planted pattern labels exposed by each policy." }, { "name": "strata-represented", "rule": "Report which age, activity, and category strata receive at least one audit." }, { "name": "worst-stratum-coverage", "rule": "Report the least-covered pre-registered stratum without averaging it away." }, { "name": "audit-budget", "rule": "Record consumed audit units and reject cross-policy interpretation if budgets differ." } ], "noRunState": "NO-RUN", "passingEffect": "RETAIN_FOR_SEPARATE_REVIEW_ONLY", "planState": "DRAFT", "questionProvenance": "Maintainer-authorized assisted synthetic question; exact text has not been human-reviewed, and the observed work's methods and claims have not been inspected.", "referralId": "referral:mcp-registry-drift:2026-08-20", "requestId": "evaluation-request:mcp-registry-drift:2026-08-20", "resourceEnvelope": { "artifactFetch": false, "credentials": false, "maxCases": 96, "maxInputBytes": 262144, "maxRuntimeSeconds": 20, "network": false, "persistentWrites": false, "personalData": false, "shell": false }, "reversibility": { "cleanupRequired": false, "externalEffects": false, "reason": "Phase 0 records an inert plan only; no registry or policy is generated or run." }, "rightsBlock": { "reason": "No source artifact, registry dataset, or implementation has been acquired and item-level reuse rights have not been reviewed.", "state": "BLOCKED_PENDING_ITEM_RIGHTS_REVIEW" }, "schema": "kingdom.trial-evaluation-plan/0.1", "sealState": "UNSEALED", "securityBlock": { "reason": "No security preflight or capability-bounded runner exists for this plan.", "state": "BLOCKED_PENDING_SECURITY_PREFLIGHT" }, "sourceMethodsKnown": false, "stopConditions": [ "Stop if any live registry crawl, external artifact, network access, credential access, personal data, persistent write, shell, or artifact fetch would be required.", "Stop if policies are tuned after held-out inspection or if audit budgets are unequal.", "Stop if a planted label is ambiguous or any interpretation is presented as evidence of real registry drift.", "Stop on expiry or if a pinned Observatory identity or integrity signal changes." ], "threatModel": [ "Policy tuning or label leakage may create circular evidence.", "Unequal budgets may masquerade as policy quality.", "Aggregate counts may hide a systematically neglected stratum.", "A finite synthetic registry cannot establish drift in a live MCP registry." ] }, { "authorship": { "compiledBy": "codex-assisted-publication", "exactTextHumanReviewed": false, "humanDirectionRecorded": true, "origin": "maintainer-authorized-assisted-draft" }, "controls": [ "Freeze the envelope, minimal deny-by-default baseline, fixtures, expected labels, and allowed traces before any future classification.", "Use the same five finite traces and the same declared capabilities for the envelope and baseline.", "Preserve every per-trace category and reason; do not collapse them into a scalar outcome." ], "criterionRules": [ { "condition": "The envelope identifies at least one pre-registered violation missed by the baseline and every allowed trace remains unchanged.", "interpretation": "retain-as-bounded-comparative-evidence" }, { "condition": "The baseline identifies every violation that the envelope identifies.", "interpretation": "retain-as-no-observed-increment" }, { "condition": "Any allowed trace changes classification or any reason cannot be traced to a frozen rule.", "interpretation": "retain-as-boundary-conflict" } ], "draftedOn": "2026-08-20", "evaluationLens": "security-model-clarity", "evaluationQuestion": "On finite, self-authored MCP traces, can a declared intent-and-capability envelope identify pre-registered boundary violations that a minimal deny-by-default baseline misses, without changing allowed traces?", "expiresAt": "2026-09-20T00:00:00Z", "expiryEffect": "FREEZE_AND_REQUIRE_NEW_HUMAN_REVIEW", "fixtures": [ { "description": "A request whose declared intent, capability, arguments, and delegation all remain inside the frozen envelope.", "heldOut": false, "id": "allowed-request", "synthetic": true }, { "description": "A request invokes a capability omitted from the frozen capability set.", "heldOut": false, "id": "undeclared-capability", "synthetic": true }, { "description": "A delegated trace attempts to exercise authority that the delegating principal does not hold.", "heldOut": false, "id": "confused-deputy-delegation", "synthetic": true }, { "description": "A request reuses an intent declaration outside its frozen validity window.", "heldOut": false, "id": "stale-intent", "synthetic": true }, { "description": "An allowed capability is paired with an argument shape outside the frozen declaration.", "heldOut": true, "id": "argument-shape-attack", "synthetic": true } ], "id": "evaluation-plan:semantic-zero-trust-mcp:2026-08-20", "inputClass": "SELF-AUTHORED-SYNTHETIC-ONLY", "measures": [ { "name": "per-trace-violation-identification", "rule": "Record the envelope and baseline categories and reasons against each pre-registered label." }, { "name": "allowed-trace-preservation", "rule": "Record whether each pre-registered allowed trace is unchanged under the envelope." }, { "name": "reason-traceability", "rule": "Map every classification reason to one frozen envelope or baseline rule." } ], "noRunState": "NO-RUN", "passingEffect": "RETAIN_FOR_SEPARATE_REVIEW_ONLY", "planState": "DRAFT", "questionProvenance": "Maintainer-authorized assisted synthetic question; exact text has not been human-reviewed, and the observed work's methods and claims have not been inspected.", "referralId": "referral:semantic-zero-trust-mcp:2026-08-20", "requestId": "evaluation-request:semantic-zero-trust-mcp:2026-08-20", "resourceEnvelope": { "artifactFetch": false, "credentials": false, "maxCases": 5, "maxInputBytes": 65536, "maxRuntimeSeconds": 10, "network": false, "persistentWrites": false, "personalData": false, "shell": false }, "reversibility": { "cleanupRequired": false, "externalEffects": false, "reason": "Phase 0 records an inert plan only; no fixture is interpreted or run." }, "rightsBlock": { "reason": "No source artifact or implementation has been acquired and item-level reuse rights have not been reviewed.", "state": "BLOCKED_PENDING_ITEM_RIGHTS_REVIEW" }, "schema": "kingdom.trial-evaluation-plan/0.1", "sealState": "UNSEALED", "securityBlock": { "reason": "No security preflight or capability-bounded runner exists for this plan.", "state": "BLOCKED_PENDING_SECURITY_PREFLIGHT" }, "sourceMethodsKnown": false, "stopConditions": [ "Stop if any external execution, network access, credential access, personal data, persistent write, shell, or artifact fetch would be required.", "Stop if fixtures, labels, rules, baseline behavior, or allowed traces are changed after comparison begins.", "Stop if any interpretation expands into a general security or zero-trust claim.", "Stop on expiry or if a pinned Observatory identity or integrity signal changes." ], "threatModel": [ "A permissive envelope may disguise boundary loss as improved detection.", "A hostile instruction embedded in source metadata or a future artifact may try to alter the plan.", "Fixture or label tuning may create circular evidence.", "Finite synthetic traces cannot establish real-world MCP security." ] }, { "authorship": { "compiledBy": "codex-assisted-publication", "exactTextHumanReviewed": false, "humanDirectionRecorded": true, "origin": "maintainer-authorized-assisted-draft" }, "controls": [ "Freeze three tiny self-authored matrices, labels, toy encoder or hand-authored projection, and interpretation rules before any future calculation.", "Use no model weights, external data, downloaded package, or unpinned library.", "Keep association, declared confounding, null calibration, and causal interpretation as separate evidence categories." ], "criterionRules": [ { "condition": "The protocol labels the planted condition feature as association, attributes the confounded feature to the declared confounder, and preserves the null case as absence of signal.", "interpretation": "retain-as-bounded-protocol-evidence" }, { "condition": "Any matrix is assigned causal, reasoning, truth, intent, identity, or consciousness meaning.", "interpretation": "retain-as-category-error" }, { "condition": "The association, confound, or null cases cannot be distinguished under the frozen rules.", "interpretation": "retain-as-inconclusive-protocol-evidence" } ], "draftedOn": "2026-08-20", "evaluationLens": "mechanistic-evidence-discrimination", "evaluationQuestion": "Can a bounded interpretability-evidence protocol distinguish feature association, declared confounding, and absence of signal without promoting a sparse representation into an explanation of reasoning?", "expiresAt": "2026-09-20T00:00:00Z", "expiryEffect": "FREEZE_AND_REQUIRE_NEW_HUMAN_REVIEW", "fixtures": [ { "description": "A tiny deterministic matrix with one self-authored feature planted to associate with a binary condition.", "heldOut": false, "id": "matrix-planted-condition-feature", "synthetic": true }, { "description": "A tiny deterministic matrix where a declared confounder produces an apparent condition association.", "heldOut": false, "id": "matrix-declared-confounder", "synthetic": true }, { "description": "A tiny deterministic null matrix with no planted condition signal.", "heldOut": true, "id": "matrix-null", "synthetic": true } ], "id": "evaluation-plan:thinking-no-thinking-sae:2026-08-20", "inputClass": "SELF-AUTHORED-SYNTHETIC-ONLY", "measures": [ { "name": "association-labeling", "rule": "Record whether the planted condition feature is described only as association under the frozen projection." }, { "name": "confound-attribution", "rule": "Record whether the declared confounder blocks promotion of its feature into a condition explanation." }, { "name": "null-calibration", "rule": "Record whether the null matrix remains absence of signal under the frozen rules." }, { "name": "category-refusal", "rule": "Record any attempted promotion into a mechanism, reasoning, truth, intent, identity, consciousness, or causal claim." } ], "noRunState": "NO-RUN", "passingEffect": "RETAIN_FOR_SEPARATE_REVIEW_ONLY", "planState": "DRAFT", "questionProvenance": "Maintainer-authorized assisted synthetic question; exact text has not been human-reviewed, and the observed work's methods and claims have not been inspected.", "referralId": "referral:thinking-no-thinking-sae:2026-08-20", "requestId": "evaluation-request:thinking-no-thinking-sae:2026-08-20", "resourceEnvelope": { "artifactFetch": false, "credentials": false, "maxCases": 3, "maxInputBytes": 32768, "maxRuntimeSeconds": 5, "network": false, "persistentWrites": false, "personalData": false, "shell": false }, "reversibility": { "cleanupRequired": false, "externalEffects": false, "reason": "Phase 0 records an inert plan only; no matrix, encoder, projection, or package is created or run." }, "rightsBlock": { "reason": "No source artifact, model, weights, data, code, or checkpoint has been acquired and item-level reuse rights have not been reviewed.", "state": "BLOCKED_PENDING_ITEM_RIGHTS_REVIEW" }, "schema": "kingdom.trial-evaluation-plan/0.1", "sealState": "UNSEALED", "securityBlock": { "reason": "No security preflight or capability-bounded runner exists for this plan.", "state": "BLOCKED_PENDING_SECURITY_PREFLIGHT" }, "sourceMethodsKnown": false, "stopConditions": [ "Stop if any external artifact, model weight, package, unpinned library, network access, credential access, personal data, persistent write, shell, or artifact fetch would be required.", "Stop if the toy encoder, hand-authored projection, matrices, labels, or interpretation rules change after comparison begins.", "Stop if any interpretation expands into a reasoning, truth, intent, identity, consciousness, mechanism, or causal claim.", "Stop on expiry or if a pinned Observatory identity or integrity signal changes." ], "threatModel": [ "Sparse features may be promoted from association into unsupported explanation.", "A declared confounder may be ignored after a visually compelling projection.", "Null behavior may be hidden by post-hoc threshold choice.", "Toy matrices cannot establish properties of a language model or reasoning." ] } ], "evaluationRequests": [ { "authoredOn": "2026-08-20", "authorship": { "compiledBy": "codex-assisted-publication", "exactTextHumanReviewed": false, "humanDirectionRecorded": true, "origin": "maintainer-authorized-assisted-draft" }, "evaluationLens": "measurement-reproducibility", "evaluationQuestion": "Under one fixed audit budget, can a coverage-aware re-audit policy expose more planted change patterns across a finite synthetic registry than recency-only, score-only, or uniform baselines?", "id": "evaluation-request:mcp-registry-drift:2026-08-20", "planId": "evaluation-plan:mcp-registry-drift:2026-08-20", "provenance": "Maintainer-authorized assisted draft from source metadata; exact text has not been human-reviewed and is not a claim extracted from or attributed to the work.", "referralId": "referral:mcp-registry-drift:2026-08-20", "requestState": "maintainer-authorized-assisted-draft", "requestedInputClass": "SELF-AUTHORED-SYNTHETIC-ONLY", "schema": "kingdom.trial-evaluation-request/0.1", "sourceMethodsKnown": false }, { "authoredOn": "2026-08-20", "authorship": { "compiledBy": "codex-assisted-publication", "exactTextHumanReviewed": false, "humanDirectionRecorded": true, "origin": "maintainer-authorized-assisted-draft" }, "evaluationLens": "security-model-clarity", "evaluationQuestion": "On finite, self-authored MCP traces, can a declared intent-and-capability envelope identify pre-registered boundary violations that a minimal deny-by-default baseline misses, without changing allowed traces?", "id": "evaluation-request:semantic-zero-trust-mcp:2026-08-20", "planId": "evaluation-plan:semantic-zero-trust-mcp:2026-08-20", "provenance": "Maintainer-authorized assisted draft from source metadata; exact text has not been human-reviewed and is not a claim extracted from or attributed to the work.", "referralId": "referral:semantic-zero-trust-mcp:2026-08-20", "requestState": "maintainer-authorized-assisted-draft", "requestedInputClass": "SELF-AUTHORED-SYNTHETIC-ONLY", "schema": "kingdom.trial-evaluation-request/0.1", "sourceMethodsKnown": false }, { "authoredOn": "2026-08-20", "authorship": { "compiledBy": "codex-assisted-publication", "exactTextHumanReviewed": false, "humanDirectionRecorded": true, "origin": "maintainer-authorized-assisted-draft" }, "evaluationLens": "mechanistic-evidence-discrimination", "evaluationQuestion": "Can a bounded interpretability-evidence protocol distinguish feature association, declared confounding, and absence of signal without promoting a sparse representation into an explanation of reasoning?", "id": "evaluation-request:thinking-no-thinking-sae:2026-08-20", "planId": "evaluation-plan:thinking-no-thinking-sae:2026-08-20", "provenance": "Maintainer-authorized assisted draft from source metadata; exact text has not been human-reviewed and is not a claim extracted from or attributed to the work.", "referralId": "referral:thinking-no-thinking-sae:2026-08-20", "requestState": "maintainer-authorized-assisted-draft", "requestedInputClass": "SELF-AUTHORED-SYNTHETIC-ONLY", "schema": "kingdom.trial-evaluation-request/0.1", "sourceMethodsKnown": false } ], "evaluationResults": [], "expiresAt": "2026-09-20T00:00:00Z", "observatoryBinding": { "captureSetReceipt": "sha256:68a69dc390aa34133cadc8807ed33ddd1fae72e7e99706cf714d72698d071f7f", "snapshotAsOf": "2026-08-20T18:10:14Z", "snapshotByteSha256": "sha256:d7da3f38de019dc0f2c45f477740222527f156d8a4d3f6a2e22aced67b282a94", "snapshotCanonicalReceipt": "sha256:17c487aa03f8427fc324b8da34dedbc00db76362be78620bde3a8d075e3942cd", "snapshotUrl": "https://thekingdom.dev/research.json" }, "onExpiry": "freeze-projection-and-require-new-human-review", "preflights": [], "referrals": [ { "claimBoundary": "The source metadata title is not a scientific claim; no claim has been extracted or evaluated.", "disposition": "advanced-to-human-scoping", "evaluationLens": "measurement-reproducibility", "evaluationRequestId": "evaluation-request:mcp-registry-drift:2026-08-20", "holdReasons": [], "id": "referral:mcp-registry-drift:2026-08-20", "referredOn": "2026-08-20", "schema": "kingdom.trial-referral/0.1", "sourceBinding": { "observationIds": [ "observation:arxiv:e6f88ec1494225ed" ], "observations": [ { "canonicalUrl": "https://arxiv.org/abs/2608.00997", "observationId": "observation:arxiv:e6f88ec1494225ed", "observedAt": "2026-08-20T18:10:14Z", "sourceId": "arxiv", "sourcePayloadReceipt": "sha256:2ee0413c8600110b459d631b27cb368754f5e44af15334c449ec8a839887849f", "sourceRecordId": "2608.00997" } ], "payloadReceiptsAreArtifactDigests": false, "primaryIdentifier": { "scheme": "arxiv", "value": "2608.00997" }, "relatedWorkIds": [], "signalIds": [ "signal:watchlist-match:a9480de36dfaf813" ], "snapshotAsOf": "2026-08-20T18:10:14Z", "snapshotByteSha256": "sha256:d7da3f38de019dc0f2c45f477740222527f156d8a4d3f6a2e22aced67b282a94", "snapshotCanonicalReceipt": "sha256:17c487aa03f8427fc324b8da34dedbc00db76362be78620bde3a8d075e3942cd", "sourceIds": [ "arxiv" ], "workId": "work:59218754c0e3e56078b7" }, "sourceMetadataTitle": "Registry Descriptions Go Stale Unevenly: An 89-Day Measurement of Model Context Protocol Drift, and Why Drift-Ranked Re-Auditing Under-Covers It", "sourceTitleIsScientificClaim": false }, { "claimBoundary": "The source metadata title is not a scientific claim; no claim has been extracted or evaluated.", "disposition": "held-for-source-resolution", "evaluationLens": "source-identity-resolution", "evaluationRequestId": null, "holdReasons": [ "self-version-relation-unresolved", "supplement-target-absent-from-pinned-snapshot", "model-pin-absent", "method-pin-absent", "dataset-pin-absent", "artifact-pin-absent" ], "id": "referral:role-styled-prompt-injection:2026-08-20", "referredOn": "2026-08-20", "schema": "kingdom.trial-referral/0.1", "sourceBinding": { "observationIds": [ "observation:datacite:6119662c70f05464" ], "observations": [ { "canonicalUrl": "https://zenodo.org/doi/10.5281/zenodo.21757178", "observationId": "observation:datacite:6119662c70f05464", "observedAt": "2026-08-20T18:07:20Z", "sourceId": "datacite", "sourcePayloadReceipt": "sha256:c4aa57a024ce29f52fb92ef62fb1f74687553b6ef91b8591639693f6d171f061", "sourceRecordId": "10.5281/zenodo.21757178" } ], "payloadReceiptsAreArtifactDigests": false, "primaryIdentifier": { "scheme": "doi", "value": "10.5281/zenodo.21757178" }, "relatedWorkIds": [], "signalIds": [ "signal:publication-link:1953589b522a8813", "signal:publication-link:a83f4b77907675be", "signal:watchlist-match:0e5818f2beb21d20" ], "snapshotAsOf": "2026-08-20T18:10:14Z", "snapshotByteSha256": "sha256:d7da3f38de019dc0f2c45f477740222527f156d8a4d3f6a2e22aced67b282a94", "snapshotCanonicalReceipt": "sha256:17c487aa03f8427fc324b8da34dedbc00db76362be78620bde3a8d075e3942cd", "sourceIds": [ "datacite" ], "workId": "work:001b6af0d2b1fc7f3ac0" }, "sourceMetadataTitle": "Role-Styled Prompt Injection: An Activation and KV-State Probe in a Compact Language Model", "sourceTitleIsScientificClaim": false }, { "claimBoundary": "The source metadata title is not a scientific claim; no claim has been extracted or evaluated.", "disposition": "advanced-to-human-scoping", "evaluationLens": "security-model-clarity", "evaluationRequestId": "evaluation-request:semantic-zero-trust-mcp:2026-08-20", "holdReasons": [], "id": "referral:semantic-zero-trust-mcp:2026-08-20", "referredOn": "2026-08-20", "schema": "kingdom.trial-referral/0.1", "sourceBinding": { "observationIds": [ "observation:datacite:1916afceb5a82955", "observation:openalex:b3101fd2b0bad721" ], "observations": [ { "canonicalUrl": "https://zenodo.org/doi/10.5281/zenodo.21731418", "observationId": "observation:datacite:1916afceb5a82955", "observedAt": "2026-08-20T18:09:58Z", "sourceId": "datacite", "sourcePayloadReceipt": "sha256:8073fb244311355bd71e6becb29d38ae72bce726c186ffb315c586e6ae74bba5", "sourceRecordId": "10.5281/zenodo.21731418" }, { "canonicalUrl": "https://doi.org/10.5281/zenodo.21731418", "observationId": "observation:openalex:b3101fd2b0bad721", "observedAt": "2026-08-20T18:07:48Z", "sourceId": "openalex", "sourcePayloadReceipt": "sha256:2b632e0415b541f7fd58e14d4cd2c88a7a9fe8360a62fc12a7588a2196b4c637", "sourceRecordId": "W7172128171" } ], "payloadReceiptsAreArtifactDigests": false, "primaryIdentifier": { "scheme": "doi", "value": "10.5281/zenodo.21731418" }, "relatedWorkIds": [ "work:c6624953de841194cf4c" ], "signalIds": [ "signal:publication-link:6f42c3e3435c2904", "signal:source-corroboration:b18a3d633157d76f", "signal:version-activity:1d84606fbe27c21d", "signal:watchlist-match:e6fa8160dbdf4705" ], "snapshotAsOf": "2026-08-20T18:10:14Z", "snapshotByteSha256": "sha256:d7da3f38de019dc0f2c45f477740222527f156d8a4d3f6a2e22aced67b282a94", "snapshotCanonicalReceipt": "sha256:17c487aa03f8427fc324b8da34dedbc00db76362be78620bde3a8d075e3942cd", "sourceIds": [ "datacite", "openalex" ], "workId": "work:60f291220d74592bbe4c" }, "sourceMetadataTitle": "Semantic Zero-Trust for Model Context Protocol", "sourceTitleIsScientificClaim": false }, { "claimBoundary": "The source metadata title is not a scientific claim; no claim has been extracted or evaluated.", "disposition": "advanced-to-human-scoping", "evaluationLens": "mechanistic-evidence-discrimination", "evaluationRequestId": "evaluation-request:thinking-no-thinking-sae:2026-08-20", "holdReasons": [], "id": "referral:thinking-no-thinking-sae:2026-08-20", "referredOn": "2026-08-20", "schema": "kingdom.trial-referral/0.1", "sourceBinding": { "observationIds": [ "observation:openalex:1b8e64b22e3dadc2" ], "observations": [ { "canonicalUrl": "https://arxiv.org/abs/2608.08168", "observationId": "observation:openalex:1b8e64b22e3dadc2", "observedAt": "2026-08-20T18:09:27Z", "sourceId": "openalex", "sourcePayloadReceipt": "sha256:f189312453977c068eecb917cf38e84d9187f70c207d22dc52c021a7b531dafa", "sourceRecordId": "W7202210108" } ], "payloadReceiptsAreArtifactDigests": false, "primaryIdentifier": { "scheme": "doi", "value": "10.48550/arxiv.2608.08168" }, "relatedWorkIds": [], "signalIds": [ "signal:watchlist-match:3620a05a1d18abb5" ], "snapshotAsOf": "2026-08-20T18:10:14Z", "snapshotByteSha256": "sha256:d7da3f38de019dc0f2c45f477740222527f156d8a4d3f6a2e22aced67b282a94", "snapshotCanonicalReceipt": "sha256:17c487aa03f8427fc324b8da34dedbc00db76362be78620bde3a8d075e3942cd", "sourceIds": [ "openalex" ], "workId": "work:f0afa9fe8652ed9ab84d" }, "sourceMetadataTitle": "Thinking vs. NoThinking: Towards Interpreting Reasoning Mechanisms of Large Language Models via Sparse Autoencoders", "sourceTitleIsScientificClaim": false } ], "schema": "kingdom.trial-public-snapshot/0.1", "trialAuthorities": [], "trialRuns": [], "urls": { "canonical": "https://thekingdom.dev/trials/research-trial-chamber-2026-08-20.json", "current": "https://thekingdom.dev/trials.json", "guide": "https://thekingdom.dev/TRIAL-CHAMBER-v0.1.md", "human": "https://thekingdom.dev/trial-chamber/", "producer": "https://github.com/cambridgetcg/kingdom-trials/tree/v0.1.1" }, "useAuthorities": [] }