EVIDENCE VAULT
Stamped IDA Evidence Packs
Redacted format examples: what ships with a Stamp, what counterparties can verify, and how status changes over time.
Use case: High-impact tool execution, including automated security testing, where unsafe defaults, scope drift, or configuration changes must fail-closed.
Start here (90 seconds)
Three actions to forward this to Procurement, Legal, or Security.
What a Stamp proves / doesn't prove
Proves (within scope)
- ✓Decision class + timestamp
- ✓Signer/authority reference
- ✓Scope boundary + expiry/evidence window
- ✓Validity state (VALID / NEEDS REFRESH / WITHDRAWN / NOT VERIFIED)
- ✓Append-only history pointer
- ✓Refresh/withdraw semantics + fail-closed verification
Does not prove
- ✗Outcome correctness or underlying truth
- ✗Certification/compliance
- ✗Raw PII/payloads, prompts, logs (excluded by default)
In disputes: Status Link = reliance state now. IDA Evidence Pack = decision-time record.
Procurement Pack includes
Status Link specimenIDA Evidence Pack specimenClause Pack / Schedule A templateSecurity controls snapshot (1 page)Data residency + subprocessors summary (1 page)SLA targets + support model (1 page)
Forward this to procurement
Send these specimens to Procurement, Legal, Security, or a counterparty. No portal access. No internal login. Just a link that fails closed.
Specimens are redacted format exemplars — not client data.
Security & privacy posture
Minimal disclosure by default (no prompts/logs/PII/payloads)
Programme-gated sealed annex (audited access trail) when legally required
Fail-closed verification (unreachable = NOT VERIFIED)
Append-only history (withdrawal ≠ erasure)
Encryption at rest (AES-256) + TLS 1.3 in transit
Hosting: US (us-east-1, us-west-2). EU/UK/AU options available for programme-scoped deployments.
Optional tamper-evident anchoring available; not required.
CLASS: VERIFY OBJECT
Status Link
The live verification URL counterparties check today.
- •Live status: VALID / NEEDS REFRESH / WITHDRAWN / NOT VERIFIED
- •Scope, expiry, signer, version in one URL
- •Official verifier: verify.goodproof.mindchill.ai
- •Unreachable = NOT VERIFIED (fail-closed)
FAIL-CLOSEDAPPEND-ONLY
How buyers use this →
Procurement analystCounterparty auditorOmbuds/tribunal reviewerAutomated gate
CLASS: EVIDENCE PACK
IDA Evidence Pack
The forwardable snapshot for procurement, disputes, and audit.
- •Time-stamped snapshot you can file, forward, and cite
- •Append-only pointer (withdrawal ≠ erasure)
- •Prompts/logs/PII excluded by default
APPEND-ONLYMINIMAL DISCLOSURE
How buyers use this →
CounselRegulatorInternal audit committeeDisputes team
CLASS: ISSUANCE
Good Proof Stamp
The core issuance record for high-impact actions.
- •Scope boundaries + evidence window
- •Refresh and withdraw triggers
- •References to Status Link + IDA Evidence Pack
VALIDFAIL-CLOSED
How buyers use this →
Contract schedule referenceTicket/runbook referenceSystem integration
CLASS: HUMAN FINALITY
Operational Guardians
Scoped human review for exceptions and disputes.
- •Conflict check + rotation protocol
- •Multi-review for high-risk lanes
- •Full audit trace
CONFLICT-CHECKEDMULTI-REVIEW
How buyers use this →
Exception lanes onlyConflict checks + rotationAudit trail required
CLASS: DELIVERY PACK
Stamp Sprint Pack
What ships in 30 days for one decision class — gated, verifiable, and ready to pilot.
- •Day 0–30 timeline + success criteria
- •Scope, workflow, triggers, checklists
- •Redacted specimen set + go/no-go memo
30-DAY1 LANE
How buyers use this →
Procurement evaluationPilot scope alignmentOnboarding team
CLASS: VERIFICATION RAILS
Verify API
Programmatic verification at scale — for gates, audits, and counterparty checks.
- •Status lookup + signed responses (programme-scoped)
- •Webhooks with idempotent event IDs
- •Access request trail (audited)
FAIL-CLOSEDSIGNED (WHEN ENABLED)
How buyers use this →
Webhook consumersAudit sampling pipelinesSigned verify responses
CLASS: HARDWARE FINALITY
Signed verify response (hardware-backed)
High-assurance verification where programmes require non-exportable signing keys.
- •Verifiable proof of status + scope at a specific time
- •Hardware-backed signature (HSM/TPM where configured)
- •Programme trust chain verifiable; no PII by default
HSM/TPMNON-EXPORTABLE
How buyers use this →
High-assurance programmes onlyInsurer/auditor verificationTamper-evident reliance
CLASS: DELEGATION AUTHORITY
Delegation Scope Specimen (Redacted)
Stamped authority object showing who/what can act, within what scope, for how long.
- •Scope, expiry, signer, revocation conditions
- •Status-Link-bound authority verification
- •Minimal disclosure by default
SCOPE-BOUNDREVOCABLE
How buyers use this →
Counterparty verificationAudit trailDelegation disputes
CLASS: SIGNED RESPONSE
Cross-Rail Verify Response (Signed, Programme-Scoped)
Verifiable status+scope at timestamp for cross-rail agent transactions.
- •Verifiable status+scope at timestamp
- •Signed response chain
- •No PII by default
SIGNEDPROGRAMME-SCOPED
How buyers use this →
Agentic commerce gatesPayment railsCross-system verification
Optional programme artefact (AI GOLD-enabled programmes only). Not a financial instrument.
DUTY-OF-CARE EVIDENCE
AI GOLD Duty-of-Care Evidence Specimen (Redacted)
What a counterparty can file. No portal access. No system login. Just a verifiable artifact.
AI GOLD Evidence Bundle
VALID
Stamp Reference
stamp_id: GP-2026-XXXX-XXX
status: VALID
verified_at: 2026-01-15T09:30:00Z
expires_at: 2026-02-14T23:59:59Z
signer: policy_signer_v3
verify_url: verify.goodproof.mindchill.ai/…
Scope
action_class: duty_of_care:high_impact
evidence_window: 30 days
Scope boundaries:
- •Hardship determination decisions within defined thresholds
- •Eligibility criteria evaluation (programme-defined)
- •Exception/override pathway validation
Duty-of-Care Evidence Bundle (categories only)
Policy basis category
Safeguarding/vulnerability considerations
Eligibility/threshold criteria
Appeal/exception route: present
"What was considered" checklist: completed
NEEDS REFRESH triggers
- • Policy/rule pack update
- • Evidence window expiry
- • Material change in evidence source
- • Model retrain affecting scope
WITHDRAWN semantics
Stop relying immediately; history remains append-only. Withdrawal ≠ erasure.
Minimal disclosure: Excluded by default: PII/PHI, prompts, logs, internal identifiers. Programme-gated access available when required (auditable trail).
This is what a counterparty can file. No portal access. No system login. Just a verifiable artifact.
Limitations: Not a financial instrument. Metrics reflect operational throughput only. No promised returns, appreciation, or guaranteed value. Programme-scoped; eligibility rules apply.
TRANSPARENCY LAYER
Coverage Report Specimen (Redacted)
Aggregate operational metrics from eligible Good Proof throughput. No per-entity disclosure.
Coverage Report
MONTHLYReporting Period
report_id: GPR-2026-01
period: 2026-01-01 → 2026-01-31
published_at: 2026-02-05T12:00:00Z
methodology: v1.2
Included Lanes Summary
InsurancePlatformsGovernment+ 4 others
X,XXX
Action counts
XX.X%
Validity retention rate
XXX
Refresh triggers fired
XX
Withdrawals issued
XX
Disputes reviewed
XXX
Guardian hours funded
Methodology Notes
- • Aggregate counts only; no per-entity or per-case disclosure
- • Test traffic and opt-out programmes excluded
- • Validity retention = Stamps remaining VALID through evidence window
- • Guardian hours = conflict-checked human finality time deployed
Aggregate transparency. No sensitive payloads. No portals. Just measurable metrics.
PDFs are great for filing. Status Links keep them current.
Common procurement questions
Ready to see it in your workflow?
Ship one fail-closed gate for one high-impact decision class in 30 days.
Start with one lane. Ship one gate in 30 days. Expand when counterparties rely on the Status Link.
Not a certification. Scope-limited verification. Acceptance depends on counterparty/programme requirements.