LISR Changelog
Every LISR score, framework revision, and methodology change is dated, versioned, and locked at publication. This changelog is the public record. Entries cannot be retroactively altered — corrections are published as new dated entries.
MetaMask Agent Wallet Scored: ACR 5.1 / MODERATE
Sixth agentic protocol scored, and the lowest risk of the six. Reached general availability 2026-08-06. Strongest programmable guardrails in the index at 1.9 — onchain spend limits, a two-tier permission model, and a mandatory per-transaction pipeline of simulation, threat scanning and MEV protection. Session key handling scores 2.1 on a trusted execution environment model where the agent holds only a request credential. Offsetting: audit log generation 8.4, quantum resistance 8.5, and multi-party approval 7.8, none of which are addressed in the published documentation. Primary failure mode recorded as plausible error — MetaMask’s own guidance notes that its optional Beast Mode removes the layer that would catch a legitimate-looking transaction the agent should not have attempted. linkmerica.com/protocols/metamask-agent-wallet/
Coverage: Agentic Payments Alliance Added to Monitoring
The Agentic Payments Alliance, convened by Rain with 26 founding members including Visa, Mastercard, Circle, Fiserv, Solana, Fireblocks, Turnkey and Chainalysis, entered Linkmerica’s monitoring record on the day of its launch. The Alliance states that its early work will include testing emerging standards for agent identity and authorization — the same subject matter Linkmerica’s agentic custody framework has assessed since June 2026. No assessment is offered here. This entry records the date monitoring began. Monitored targets: 24.
Research: The Purchase Record — Customer Data Retention as a Custody Risk Input
Eleventh research brief. Three documented breach cases — Ledger 2020, Trezor and SafePal in August 2026 — and an audit of what all eight scored and pipeline manufacturers publish about customer data retention. Five state a numeric period; three do not. The range runs from 30 days to ten years. The brief states plainly that Trezor’s disclosed 90-day retention control measurably bounded its own breach, publishes Linkmerica’s limitation that this assesses disclosed policy rather than verified practice, and records that LISR v1.0’s six categories do not cover customer data retention — a framework question now under review after three of seven scored wallets experienced a material event in this area. linkmerica.com/research/the-purchase-record/
Coldcard Review — Day 14, Still Not Scored
Coinkite has stated it will publish a detailed technical post-mortem once able, and its internal review remains open. Linkmerica continues to wait on that document rather than scoring against an analysis the manufacturer describes as preliminary. New to the notice: Bitcoin developer James O’Beirne’s public account that he raised the randomness code with Coinkite in May 2025, fourteen months before the exploit surfaced. Reported as his account, not independently verified. If it holds, vendor responsiveness to a reported concern is an assurance property in its own right. linkmerica.com/wallets/coldcard/
Research: A Moving Target — Hardware Custody and the Post-Quantum Migration Problem
Tenth research brief. Connects two findings published three months apart: Project Eleven’s May analysis of why immutable bootloaders prevent hardware wallets from migrating boot verification to post-quantum cryptography, and the July withdrawal of HAWK from NIST consideration after an AI system reduced its expected attack cost in roughly sixty hours. Hardware custody commits to primitives for years; algorithm confidence has demonstrated it can change in days. Reframes crypto-agility as the property that absorbs that risk, and states plainly that the framework’s regulatory alignment category may under-weight it. No deployed system is affected and the finalised NIST standards are unaffected. linkmerica.com/research/post-quantum-moving-target/
Coldcard Review — Status Update, Still Not Scored
Device-level facts have largely settled: root cause confirmed, affected firmware enumerated by model and unchanged since August 1, fixed firmware released for every affected model. The outstanding item is Coinkite’s own formal technical review, which the company states is forthcoming and describes its current analysis as preliminary. Linkmerica is waiting on that specific document rather than on the incident ending. Reported losses continue to rise but describe the scale of the event rather than a property of the device, and are not an input to a LISR score. linkmerica.com/wallets/coldcard/
Ledger QRR Corrected: 6.8/HIGH → 5.4/MODERATE
A correction to Linkmerica’s own work. Two category rationales asserted that Ledger had announced no production post-quantum implementation and published no FIPS alignment statement. Ledger’s Embedded OS team published an ML-KEM (FIPS 203) and ML-DSA (FIPS 204) implementation guide in June 2026, before the original scoring date. Credit is partial and the tier change is capped: the implementation sits in the SDK, available to applications, rather than protecting the boot chain, firmware verification, or transaction signing. This is the third wallet in three weeks whose quantum score was assessed on an absence that manufacturer documentation contradicted. linkmerica.com/corrections/
Methodology: What a Tier Does Not Mean
Standing methodology statement added. LISR assesses documented and disclosed controls; it cannot detect undisclosed implementation defects. A LOW tier means the disclosed control set is strong, not that the implementation is proven correct. This applies to every tier. Referenced from all score pages. Also published: Passport Prime Coldcard lineage check (no inheritance found on the public timeline; code-level review not performed), and two corrections to the Passport Prime page — an analyst note stating 4.2 where the locked score is 4.1, and superseded QuantumLink reasoning predating the July 29 correction. linkmerica.com/methodology/
Coldcard Review Notice Updated — Scope and Scale Both Expanded
The Coldcard review notice published July 31 has been updated. A second wave of sweeps has been identified by Galaxy Research, and the scope of affected devices has widened rather than narrowed since initial disclosure. Loss figures have been revised upward repeatedly within 48 hours. Coldcard still carries no LISR score. Had one been published on July 31 it would already require revision — which is the reason it was withheld. linkmerica.com/wallets/coldcard/
BitBox02 Scored: LISR 3.5 / LOW · QRR 7.6 / HIGH
Seventh wallet in the index, and the first to score in the LOW custody risk tier. BitBox02 pairs a secure element with fully open, reproducibly-built firmware independently verified by WalletScrutiny — a combination no other scored wallet currently offers. Offsetting factors: a manufacturer-disclosed OLED power-consumption side channel that remains unmitigated on this model, closed-silicon dependency on the ATECC608B, and a threat model that explicitly excludes tampered devices. Quantum readiness scores 7.6/HIGH: no published post-quantum position, roadmap, or NIST alignment of any kind, partially offset by a proven signed and reproducible firmware update path. SCOPE: covers the original USB-only BitBox02. The BitBox02 Nova is a separate concurrently-sold device using a different secure chip and adding a Bluetooth surface; it is not covered by this score. linkmerica.com/wallets/bitbox02/
Coverage: Coldcard (Coinkite) — Review Opened, Not Scored
Review opened following the reported movement of approximately 594 BTC from roughly 500 single-signature addresses and Coinkite's subsequent security advisory on seed entropy generation. Coldcard does not carry a LISR score. Score withheld pending stabilisation of the evidence base — Coinkite's investigation is ongoing and the scope of affected devices has already been revised. Coinkite has not confirmed the entropy issue caused the observed movement of funds. linkmerica.com/wallets/coldcard/
Research: Tempo MPP — First Tier 2 Protocol Coverage
Ninth research brief, and the first covering a Tier 2 monitoring target. Core finding: Visa operates a production validator node on Tempo's permissioned institutional consortium — a different order of commitment from consortium membership, and a different custody question from the centralised key directory Linkmerica scored in Visa TAP. No ACR score assigned; this is qualitative analysis. linkmerica.com/research/tempo-mpp/
Foundation Passport Prime QRR Corrected: 7.3 → 6.3 (tier unchanged, HIGH)
A correction to Linkmerica's own work, not a rescore. Two category rationales in the 2026-07-12 score incorrectly asserted that Foundation had not publicly named a NIST-standardized post-quantum algorithm; Foundation named ML-KEM (FIPS 203) seven weeks earlier. The score's central judgment is unchanged and correct: the feature protects the wireless transport layer, not transaction signing or the boot chain. Risk tier remains HIGH. Original file locked and unaltered. Full detail: linkmerica.com/corrections/
Tier 1 Agentic Protocol Coverage Complete — Ripple XRPL 6.2/HIGH + Synthesis Brief
Ripple XRPL AI Starter Kit scored 6.2/HIGH, completing independent custody assessment of all five Tier 1 agentic payment protocols within 90 days of the ACR framework's founding. Ripple is the first protocol of the five to explicitly disclose multi-signature support for agent transactions. Score carries MODERATE confidence — the only Tier 1 score not rated HIGH — reflecting Phase 1 immaturity and a secondary-source evidence base; treated as provisional. Published alongside the synthesis brief "Five Ways an Agentic Protocol Fails Custody." Full cohort: Visa TAP 7.2, AP4M 6.6, Google AP2 6.4, Ripple XRPL 6.2, Coinbase AgentKit 6.0. linkmerica.com/research/agentic-protocol-failure-modes/
Trezor QRR Rescored: 7.8/HIGH → 4.4/MODERATE — Device Generation Correction
Linkmerica's first risk-decreasing QRR rescore. The Trezor Safe 7 shipped production post-quantum cryptography for its boot and firmware-update chain (hybrid classical + post-quantum signature verification, factory-immutable boardloader) — the first hardware wallet in the industry to do so. This is not the same as quantum-safe custody: transaction-signing cryptography remains classical pending blockchain-level PQC adoption. This rescore also corrects a research gap in the original June 28 score, which incorrectly stated no public Trezor quantum-readiness statement existed; SatoshiLabs had published one in March 2026. Full correction note: /srv/linkmerica/scoring/corrections/correction_trezor_qrr_2026-07-24.md. The original June 28 file remains locked and unaltered — this is an additive, dated correction, not a retroactive edit. linkmerica.com/wallets/trezor/
LISR-ACR-v1.0 — Fourth Agentic Protocol Score: Coinbase AgentKit + x402 — 6.0 / MODERATE
Fourth independent agentic protocol custody score — the first to score below HIGH. Only scored protocol to disclose functioning session-key and guardrail mechanisms (Agentic Wallets), scoring materially lower on those categories. Distinct risk: official documentation explicitly warns against production use of the public facilitator, and the July 14 Linux Foundation governance transfer (40+ members including direct competitors) introduces novel consortium-governance risk. linkmerica.com/protocols/coinbase-agentkit/
Research: Coinbase AgentKit + x402 Custody Risk Brief Published
Seventh research brief. Core finding: x402 transferred from Coinbase to Linux Foundation governance July 14, 2026, with 40+ Premier Members including direct competitors Visa, Mastercard, and Google. ACR score forthcoming July 23. linkmerica.com/research/coinbase-agentkit/
LISR-ACR-v1.0 — Third Agentic Protocol Score: Visa TAP — 7.2 / HIGH
Third independent agentic protocol custody score, and the highest of the three. Core finding: a documented, named architectural single point of failure in Visa's centralized agent key directory — not an absence-of-disclosure risk like AP4M and AP2, but a structural design gap acknowledged in TAP's own original specification. linkmerica.com/protocols/visa-tap/
Research: Visa TAP Custody Risk Brief Published
Sixth research brief — independent custody risk assessment of Visa's Trusted Agent Protocol. Core finding: structural single-point-of-failure in the centralized key directory architecture; federated model unimplemented with no published timeline. ACR score forthcoming July 16. linkmerica.com/research/visa-tap/
LISR-ACR-v1.0 — Second Agentic Protocol Score: Google AP2 — 6.4 / HIGH
Second independent agentic protocol custody score. Scored against Linkmerica's dated monitoring record (June 15 – July 12). Qualitatively different risk profile than AP4M: AP2's risks are visible and research-validated (demonstrated prompt injection attack, no native mandate revocation) rather than undisclosed. linkmerica.com/protocols/google-ap2/
New Wallet: Foundation Passport Prime — LISR 4.1/MODERATE, QRR 7.3/HIGH
Sixth wallet added to the LISR index. First device with an independent published hardware security audit (Keylabs) and the first with designed-in wireless connectivity. QRR assessment declined to credit the manufacturer's unverified "post-quantum encrypted" marketing claim absent named algorithms or independent validation. linkmerica.com/wallets/passport-prime/
Tangem Rescored — 5.8/MODERATE → 6.8/HIGH
First tier-changing rescore since founding. Driven by Ledger Donjon's July 10, 2026 disclosure of an unpatchable laser fault-injection vulnerability enabling password reset and fund drainage. No firmware update mechanism exists to remediate. linkmerica.com/wallets/tangem/
LISR-ACR-v1.0 — First Agentic Protocol Score: Mastercard AP4M — 6.9 / HIGH
The first independent agentic protocol custody score in existence. Scored against Linkmerica's dated monitoring record (June 11–28). Primary finding: absence of public disclosure on session key scoping, guardrails, and multi-party approval. linkmerica.com/protocols/ap4m/
Research: Google AP2 Custody Risk Brief Published
Fifth research brief — nine months of AP2 adoption data and the authorization custody risk category. linkmerica.com/research/google-ap2/
Research: AP4M 30-Day Follow-On Published
Fourth research brief published — the first institutional research document built entirely on a dated monitoring record (June 11–16 sweeps). linkmerica.com/research/ap4m-30-days/
LISR-QRR-v1.0 — First Quantum Resistance Scores Published
First independent quantum readiness scores for hardware wallets: Ledger 6.8/HIGH, Trezor 7.8/HIGH, Tangem 9.1/CRITICAL, SafePal 8.3/CRITICAL, ELLIPAL 8.7/CRITICAL. Published to all wallet pages.
LISR v1.0 Score: ELLIPAL Titan — 6.4 / HIGH
Fifth wallet scored under LISR v1.0. First air-gapped wallet in the index. Score locked at publication, filed under period 2026-07.
Monitor Expansion: 22 Targets — EO Compliance Milestones Added
Five federal EO compliance milestone targets added to the agentic monitor following Executive Orders 14412 and 14413 (June 22, 2026). Watch-list changelog entry 5.
Research: Quantum EO Brief Published
Analysis of EO 14412/14413 implications for hardware wallet custody. Linkmerica published independent quantum research June 15 — one week before federal policy validated the thesis.
LISR-ACR-v1.0 — Agentic Custody Readiness Dimension Scaffolded
Six-category framework for rating agentic protocol custody risk. First protocol scoring: AP4M, July 2026. Framework version-stamped and locked.
LISR-QRR-v1.0 — Quantum Resistance Readiness Dimension Scaffolded
Five-category quantum readiness framework version-stamped following compressed Q-Day timeline research (March 2026).
Research: Quantum Readiness Gap Published
First independent institutional quantum readiness assessment of hardware wallet custody infrastructure. Published one week before EO 14412/14413.
Research: AP4M Custody Risk Analysis Published
First independent institutional custody risk analysis of Mastercard AP4M — published the day after launch. Agentic infrastructure monitoring begins: 13 targets.
LISR v1.0 Scores: Ledger, Trezor, Tangem, SafePal
First four LISR v1.0 scores published and locked: Ledger 4.8/MODERATE, Trezor 4.7/MODERATE, Tangem 5.8/MODERATE, SafePal 6.4/HIGH.
Linkmerica Founded — LISR Framework Established
Linkmerica (CASPO LLC) founded Presidents Day 2026. LISR v1.0 framework scaffolded. Versioned scoring record begins.