Worked Example - Project Contract

{
 "documentType": "project_contract",
 "schemaVersion": "0.1.0",
 "contractId": "example-grounded-analysis",
 "contractVersion": "0.1.0",
 "projectName": "Example Evidence-Grounded Analysis",
 "description": "Synthetic example showing positive proof, calibrated conclusions, and independent annotation auditing.",
 "status": "approved",
 "createdAt": "2026-07-15T12:00:00Z",
 "effectiveFrom": "2026-07-15T12:00:00Z",
 "owners": [
 {
 "actorId": "owner-001",
 "role": "project_owner"
 },
 {
 "actorId": "expert-001",
 "role": "domain_expert"
 }
 ],
 "instructionDocuments": [
 {
 "instructionId": "instructions-main",
 "title": "Main project instructions",
 "authority": "primary",
 "content": "Use only the supplied evidence. State the strongest conclusion supported by the evidence. Distinguish association from causation, address both supplied studies, and cite the evidence for every material factual claim. Do not force a binary conclusion when the evidence is inconclusive.",
 "contentHash": "sha256:adca02e1bb85249a185e0460d27f4e9a28de241a2f835dce671cacd15a858495",
 "precedenceRank": 1
 }
 ],
 "sourcePolicy": {
 "worldMode": "closed_world",
 "parametricKnowledgePolicy": "never_evidence",
 "externalRetrievalAllowed": false,
 "defaultCitationRequirement": "material_claims",
 "snapshotRequired": true,
 "conflictPolicy": "Report material conflict or uncertainty; do not silently choose the more convenient source.",
 "freshnessRules": [],
 "evidenceClasses": [
 {
 "classId": "supplied-study",
 "name": "Supplied study excerpt",
 "description": "Evidence included with the task and permitted for all factual judgments.",
 "permitted": true,
 "authorityRank": 3
 }
 ]
 },
 "criteria": [
 {
 "criterionId": "C1-direct-response",
 "name": "Directly answers the question",
 "requirement": "State the strongest conclusion supported by the evidence about whether X is the primary cause.",
 "intent": "The response must resolve the requested issue without inventing certainty.",
 "criterionType": "required_presence",
 "scope": "attempt",
 "subjectivity": "bounded_judgment",
 "instructionRefs": [
 {
 "instructionId": "instructions-main",
 "locator": "sentence 2",
 "quotedText": "State the strongest conclusion supported by the evidence."
 },
 {
 "instructionId": "instructions-main",
 "locator": "sentence 4",
 "quotedText": "Do not force a binary conclusion when the evidence is inconclusive."
 }
 ],
 "applicability": {
 "ruleText": "Always applies.",
 "condition": {
 "kind": "always"
 }
 },
 "proofPolicy": {
 "positiveProofRequired": true,
 "positiveProofStandard": "The attempt states an explicit conclusion that answers the primary-cause question at the confidence level supported by the evidence.",
 "failureCondition": "The attempt does not state a conclusion, answers a different question, or forces a conclusion stronger than the evidence permits.",
 "omissionCondition": "No conclusion about primary causation is present.",
 "passOnNoViolation": false,
 "allowPartialCredit": true,
 "allowCompensation": false
 },
 "evidencePolicy": {
 "allowedEvidenceClassIds": [
 "supplied-study"
 ],
 "minimumAuthorityRank": 3,
 "citationRequired": false,
 "externalVerificationMode": "forbidden",
 "deterministicValidatorIds": []
 },
 "anchors": [
 {
 "anchorId": "C1-pass-anchor",
 "label": "Calibrated direct answer",
 "example": "The evidence does not establish that X is the primary cause.",
 "expectedVerdict": "met",
 "notes": "A non-binary answer is direct when the evidence is inconclusive."
 },
 {
 "anchorId": "C1-fail-anchor",
 "label": "No answer",
 "example": "The studies have several methodological differences.",
 "expectedVerdict": "not_met",
 "notes": "Describes evidence without answering the causal question."
 }
 ],
 "validAlternatives": [
 "A qualified no: the evidence does not establish primacy.",
 "An explicitly uncertain conclusion supported by both studies."
 ],
 "exceptions": [],
 "dependencies": {
 "requiredBeforeCriterionIds": [],
 "impliesCriterionIds": []
 },
 "precedence": {
 "overridesCriterionIds": [],
 "overriddenByCriterionIds": []
 },
 "defaultSeverity": "major",
 "defaultDecisionEffect": "rework",
 "tags": [
 "core",
 "answerability",
 "positive-proof"
 ]
 },
 {
 "criterionId": "C2-grounding",
 "name": "Material claims are grounded",
 "requirement": "Every material factual claim must be supported by permitted supplied evidence.",
 "intent": "Prevent unsupported facts and citation laundering.",
 "criterionType": "grounding",
 "scope": "claim",
 "subjectivity": "objective",
 "instructionRefs": [
 {
 "instructionId": "instructions-main",
 "locator": "sentence 1",
 "quotedText": "Use only the supplied evidence."
 },
 {
 "instructionId": "instructions-main",
 "locator": "sentence 3",
 "quotedText": "Distinguish association from causation, address both supplied studies, and cite the evidence for every material factual claim."
 }
 ],
 "applicability": {
 "ruleText": "Applies to all material factual claims.",
 "condition": {
 "kind": "always"
 }
 },
 "proofPolicy": {
 "positiveProofRequired": true,
 "positiveProofStandard": "Each material claim is linked to evidence that directly entails or reasonably supports it.",
 "failureCondition": "A central or supporting factual claim lacks permitted evidence or conflicts with it.",
 "omissionCondition": "A required citation or evidence link for a material claim is absent.",
 "passOnNoViolation": false,
 "allowPartialCredit": true,
 "allowCompensation": false
 },
 "evidencePolicy": {
 "allowedEvidenceClassIds": [
 "supplied-study"
 ],
 "minimumAuthorityRank": 3,
 "citationRequired": true,
 "externalVerificationMode": "forbidden",
 "deterministicValidatorIds": [
 "citation-id-validator"
 ]
 },
 "anchors": [
 {
 "anchorId": "C2-pass-anchor",
 "label": "Claim-specific support",
 "example": "E1 reports an 18% observational difference and notes self-selection.",
 "expectedVerdict": "met",
 "notes": "The claim matches the supplied excerpt."
 },
 {
 "anchorId": "C2-fail-anchor",
 "label": "Unsupported strengthening",
 "example": "E1 proves that X caused the improvement.",
 "expectedVerdict": "not_met",
 "notes": "The source explicitly disclaims causal identification."
 }
 ],
 "validAlternatives": [],
 "exceptions": [
 "Common connective language that adds no factual content does not require a citation."
 ],
 "dependencies": {
 "requiredBeforeCriterionIds": [],
 "impliesCriterionIds": []
 },
 "precedence": {
 "overridesCriterionIds": [],
 "overriddenByCriterionIds": []
 },
 "defaultSeverity": "major",
 "defaultDecisionEffect": "rework",
 "tags": [
 "grounding",
 "evidence",
 "hard-quality"
 ]
 },
 {
 "criterionId": "C3-warrant",
 "name": "Conclusion strength matches evidence",
 "requirement": "Distinguish association from causation and do not assert primacy unless the evidence warrants it.",
 "intent": "Assess whether the inference, not merely the cited facts, is justified.",
 "criterionType": "inferential_warrant",
 "scope": "claim",
 "subjectivity": "bounded_judgment",
 "instructionRefs": [
 {
 "instructionId": "instructions-main",
 "locator": "sentence 2",
 "quotedText": "State the strongest conclusion supported by the evidence."
 },
 {
 "instructionId": "instructions-main",
 "locator": "sentence 3",
 "quotedText": "Distinguish association from causation, address both supplied studies, and cite the evidence for every material factual claim."
 }
 ],
 "applicability": {
 "ruleText": "Applies to causal and comparative conclusions.",
 "condition": {
 "kind": "always"
 }
 },
 "proofPolicy": {
 "positiveProofRequired": true,
 "positiveProofStandard": "The response explicitly calibrates its causal conclusion to study design, uncertainty, and counterevidence.",
 "failureCondition": "The response infers causation or primacy from evidence that supports only association or an uncertain effect.",
 "omissionCondition": "The response fails to distinguish observational association from causal evidence.",
 "passOnNoViolation": false,
 "allowPartialCredit": true,
 "allowCompensation": false
 },
 "evidencePolicy": {
 "allowedEvidenceClassIds": [
 "supplied-study"
 ],
 "minimumAuthorityRank": 3,
 "citationRequired": true,
 "externalVerificationMode": "forbidden",
 "deterministicValidatorIds": []
 },
 "anchors": [
 {
 "anchorId": "C3-pass-anchor",
 "label": "Proportionate inference",
 "example": "X may contribute, but the evidence does not establish that it is the primary cause.",
 "expectedVerdict": "met",
 "notes": "Uses appropriate uncertainty and distinguishes contribution from primacy."
 },
 {
 "anchorId": "C3-fail-anchor",
 "label": "Causal overreach",
 "example": "The 18% difference proves X is the primary cause.",
 "expectedVerdict": "not_met",
 "notes": "Treats observational association as proof of causation and primacy."
 }
 ],
 "validAlternatives": [
 "A cautious conclusion that evidence is insufficient to determine primacy."
 ],
 "exceptions": [],
 "dependencies": {
 "requiredBeforeCriterionIds": [
 "C2-grounding"
 ],
 "impliesCriterionIds": []
 },
 "precedence": {
 "overridesCriterionIds": [],
 "overriddenByCriterionIds": []
 },
 "defaultSeverity": "major",
 "defaultDecisionEffect": "rework",
 "tags": [
 "reasoning",
 "warrant",
 "hard-quality"
 ]
 },
 {
 "criterionId": "C4-completeness",
 "name": "Addresses both studies",
 "requirement": "Use and synthesize both E1 and E2.",
 "intent": "Prevent selective use of convenient evidence.",
 "criterionType": "completeness",
 "scope": "attempt",
 "subjectivity": "objective",
 "instructionRefs": [
 {
 "instructionId": "instructions-main",
 "locator": "sentence 3",
 "quotedText": "Distinguish association from causation, address both supplied studies, and cite the evidence for every material factual claim."
 }
 ],
 "applicability": {
 "ruleText": "Applies when both studies are supplied.",
 "condition": {
 "kind": "always"
 }
 },
 "proofPolicy": {
 "positiveProofRequired": true,
 "positiveProofStandard": "The attempt materially describes the relevant contribution and limitation of E1 and E2.",
 "failureCondition": "One study is ignored or materially misrepresented.",
 "omissionCondition": "No substantive treatment of E1 or E2 appears.",
 "passOnNoViolation": false,
 "allowPartialCredit": true,
 "allowCompensation": false
 },
 "evidencePolicy": {
 "allowedEvidenceClassIds": [
 "supplied-study"
 ],
 "minimumAuthorityRank": 3,
 "citationRequired": true,
 "externalVerificationMode": "forbidden",
 "deterministicValidatorIds": [
 "required-evidence-id-validator"
 ]
 },
 "anchors": [
 {
 "anchorId": "C4-pass-anchor",
 "label": "Both sources synthesized",
 "example": "E1 shows an association but cannot identify causation; E2 is randomized but inconclusive.",
 "expectedVerdict": "met",
 "notes": "Captures the evidentiary role of both studies."
 }
 ],
 "validAlternatives": [],
 "exceptions": [],
 "dependencies": {
 "requiredBeforeCriterionIds": [],
 "impliesCriterionIds": []
 },
 "precedence": {
 "overridesCriterionIds": [],
 "overriddenByCriterionIds": []
 },
 "defaultSeverity": "major",
 "defaultDecisionEffect": "rework",
 "tags": [
 "completeness",
 "positive-proof"
 ]
 },
 {
 "criterionId": "C5-annotation-fidelity",
 "name": "Reviewer label and rationale faithfully evaluate the attempt",
 "requirement": "The reviewer must apply the written criteria, cite relevant evidence, and assign proportionate severity.",
 "intent": "Audit the human annotation separately from the attempt.",
 "criterionType": "annotation_fidelity",
 "scope": "annotation",
 "subjectivity": "bounded_judgment",
 "instructionRefs": [
 {
 "instructionId": "instructions-main",
 "locator": "entire document",
 "quotedText": "Use only the supplied evidence. State the strongest conclusion supported by the evidence. Distinguish association from causation, address both supplied studies, and cite the evidence for every material factual claim. Do not force a binary conclusion when the evidence is inconclusive."
 }
 ],
 "applicability": {
 "ruleText": "Applies only when a human review is present.",
 "condition": {
 "kind": "predicate",
 "fieldPath": "case.humanReview",
 "operator": "exists",
 "value": true
 }
 },
 "proofPolicy": {
 "positiveProofRequired": true,
 "positiveProofStandard": "The reviewer label, criterion attribution, evidence, and severity are mutually consistent and aligned with the contract.",
 "failureCondition": "The review applies an unstated preference, misattributes the issue, cites irrelevant evidence, or assigns an unsupported disposition.",
 "omissionCondition": "The rationale omits the decisive criterion or evidence.",
 "passOnNoViolation": false,
 "allowPartialCredit": true,
 "allowCompensation": false
 },
 "evidencePolicy": {
 "allowedEvidenceClassIds": [
 "supplied-study"
 ],
 "minimumAuthorityRank": 3,
 "citationRequired": false,
 "externalVerificationMode": "forbidden",
 "deterministicValidatorIds": []
 },
 "anchors": [
 {
 "anchorId": "C5-fail-anchor",
 "label": "Unstated binary preference",
 "example": "Fail because the answer did not choose yes or no, even though the instructions allow an inconclusive conclusion.",
 "expectedVerdict": "not_met",
 "notes": "The reviewer substitutes a preference that contradicts the written contract."
 }
 ],
 "validAlternatives": [],
 "exceptions": [],
 "dependencies": {
 "requiredBeforeCriterionIds": [
 "C1-direct-response"
 ],
 "impliesCriterionIds": []
 },
 "precedence": {
 "overridesCriterionIds": [],
 "overriddenByCriterionIds": []
 },
 "defaultSeverity": "major",
 "defaultDecisionEffect": "accept_with_note",
 "tags": [
 "annotation",
 "reviewer-audit"
 ]
 }
 ],
 "decisionPolicy": {
 "policyVersion": "0.1.0",
 "allowedActions": [
 "accept",
 "accept_with_notes",
 "rework",
 "reject",
 "escalate"
 ],
 "rules": [
 {
 "ruleId": "R1-critical-hard-failure",
 "priority": 1,
 "description": "Reject a calibrated critical hard-quality failure.",
 "match": {
 "criterionTags": [
 "hard-quality"
 ],
 "verdicts": [
 "not_met"
 ],
 "severities": [
 "critical"
 ],
 "minimumCalibratedConfidence": 0.95
 },
 "action": "reject",
 "terminal": true,
 "explanationTemplate": "A critical hard-quality criterion failed at high calibrated confidence.",
 "target": "attempt"
 },
 {
 "ruleId": "R2-major-repairable-failure",
 "priority": 2,
 "description": "Request rework for a clear major, repairable failure.",
 "match": {
 "verdicts": [
 "not_met"
 ],
 "severities": [
 "major"
 ],
 "minimumCalibratedConfidence": 0.85
 },
 "action": "rework",
 "terminal": false,
 "explanationTemplate": "At least one major criterion is not met and has a specific repair.",
 "target": "attempt"
 },
 {
 "ruleId": "R3-decision-changing-uncertainty",
 "priority": 3,
 "description": "Escalate unresolved findings capable of changing the final action.",
 "match": {
 "hasUnresolvedFindings": true
 },
 "action": "escalate",
 "terminal": false,
 "explanationTemplate": "A decision-changing issue remains unresolved.",
 "target": "attempt"
 },
 {
 "ruleId": "R4-accept-clean",
 "priority": 10,
 "description": "Accept when all applicable attempt-quality criteria are met and no major unresolved issue remains.",
 "match": {
 "criterionIds": [
 "C1-direct-response",
 "C2-grounding",
 "C3-warrant",
 "C4-completeness"
 ],
 "verdicts": [
 "met"
 ],
 "minimumCalibratedConfidence": 0.85,
 "hasUnresolvedFindings": false
 },
 "action": "accept",
 "terminal": true,
 "explanationTemplate": "All applicable attempt-quality criteria are affirmatively satisfied.",
 "target": "attempt"
 },
 {
 "target": "annotation",
 "ruleId": "R5-annotation-major-failure",
 "priority": 1,
 "description": "Request correction when annotation fidelity has a clear major failure.",
 "match": {
 "criterionIds": [
 "C5-annotation-fidelity"
 ],
 "verdicts": [
 "not_met"
 ],
 "severities": [
 "major"
 ],
 "minimumCalibratedConfidence": 0.85
 },
 "action": "rework",
 "terminal": true,
 "explanationTemplate": "The human annotation is not faithful to the attempt and contract."
 },
 {
 "target": "annotation",
 "ruleId": "R6-annotation-clean",
 "priority": 10,
 "description": "Accept a faithful human annotation.",
 "match": {
 "criterionIds": [
 "C5-annotation-fidelity"
 ],
 "verdicts": [
 "met"
 ],
 "minimumCalibratedConfidence": 0.85,
 "hasUnresolvedFindings": false
 },
 "action": "accept",
 "terminal": true,
 "explanationTemplate": "The annotation is faithful and adequately supported."
 },
 {
 "target": "composite_case",
 "ruleId": "R7-composite-annotation-rework",
 "priority": 1,
 "description": "Rework the composite datum when the attempt is acceptable but its required annotation is not.",
 "match": {
 "criterionIds": [
 "C5-annotation-fidelity"
 ],
 "verdicts": [
 "not_met"
 ],
 "severities": [
 "major"
 ],
 "minimumCalibratedConfidence": 0.85
 },
 "action": "rework",
 "terminal": true,
 "explanationTemplate": "The attempt can be retained, but the annotation must be corrected before the composite datum is admitted."
 }
 ],
 "defaultAction": "escalate",
 "tieBreakOrder": [
 "reject",
 "escalate",
 "rework",
 "accept_with_notes",
 "accept"
 ],
 "confidencePolicy": {
 "requiresCalibrationForActions": [
 "accept",
 "reject"
 ],
 "uncalibratedAction": "escalate",
 "minimumCalibrationCasesPerSlice": 50
 }
 },
 "humanReviewPolicy": {
 "maxInteractionsPerCase": 1,
 "onlyAskIfDecisionChanging": true,
 "hideProvisionalDecision": true,
 "requireAmbiguityOption": true,
 "allowedQuestionTypes": [
 "applicability",
 "instruction_precedence",
 "evidence_authority",
 "inference_permission",
 "severity",
 "valid_alternative",
 "rubric_ambiguity",
 "domain_expertise"
 ],
 "expertiseRoutingRules": [
 "Route causal-inference disputes to a reviewer qualified in study design."
 ]
 },
 "feedbackPolicy": {
 "audiences": [
 "attempter",
 "reviewer",
 "project_owner"
 ],
 "maxItemsPerAudience": 3,
 "includeValidatedStrengths": true,
 "validateRepairs": true,
 "priorityOrder": [
 "decisive",
 "critical",
 "major",
 "minor",
 "strength"
 ]
 },
 "securityPolicy": {
 "treatCandidateContentAsUntrusted": true,
 "ignoreEmbeddedEvaluatorInstructions": true,
 "requireStructuredOutputs": true,
 "allowCandidateToOverrideRubric": false,
 "redactionRules": [
 "Remove direct personal identifiers from reviewer analytics exports."
 ]
 },
 "auditPolicy": {
 "retainExactSourceSnapshots": true,
 "retainPromptVersions": true,
 "retainModelRunMetadata": true,
 "retainHumanResolutions": true,
 "replayRequired": true,
 "retentionDays": 365
 },
 "metadata": {
 "example": true
 },
 "decisionTargets": [
 "attempt",
 "annotation",
 "composite_case"
 ]
}