Tracker: SPEC v0.2 review findings (June 2026)

[iret77]

Umbrella issue for the June 2026 review of SPEC.md v0.2. All findings carry the spec-review label; severity is on each issue.

Critical (implementation blockers / security)

• §6.1/§6.3: Signature input and hash coverage of proof[] are undefined #4 — §6.1/§6.3: Signature input and hash coverage of proof[] undefined (incl. memorandum↔§8.3 contradiction)
• §7.4: N-party accepts conflict with the linear hash chain #5 — §7.4: N-party accepts conflict with the linear hash chain
• §6.2.1/§13: Uncontested enforcement requires contract-readable anchor state — missing DLT requirement #6 — §6.2.1/§13: Uncontested enforcement needs contract-readable anchor state (missing §13.1 requirement; Hedera/HCS inconsistency)
• §6.5: PQC-gated status mutation is impractical to enforce on-chain as specified #7 — §6.5: PQC-gated status mutation impractical to enforce on-chain
• §5.3: Mandate revocation enables TOCTOU/repudiation via a mutable HTTPS status list #8 — §5.3: Mandate revocation TOCTOU via mutable status list
• §8.3/§9.2: VRF witness/arbiter selection is grindable #9 — §8.3/§9.2: VRF witness/arbiter selection grindable
• §9.4: No evidence window or ruling deadline — escrow can be stranded in CHALLENGED #10 — §9.4: No evidence window / ruling deadline (escrow stranded in CHALLENGED)

High

• §7.2/§7.4: execute and escrow funding are not atomic #11 — §7.2/§7.4: execute and escrow funding not atomic
• §9.4/§6.2.1: Asserter bond is SHOULD-only — false assertions are free (and the two sections contradict each other) #12 — §9.4/§6.2.1: Asserter bond SHOULD-only; sections contradict each other
• §9.3/§7.4: Permissionless anchoring enables thread pollution and dispute-freeze griefing #13 — §9.3/§7.4: Permissionless anchoring → thread pollution & dispute-freeze griefing
• §7.4/§10: No mutual-settlement fast path — every payment waits out the challenge window #14 — §7.4/§10: No mutual-settlement fast path (challenge window on every happy path)
• §7: No amendment or mutual rescission after ACCEPTED/EXECUTED #15 — §7: No amendment or mutual rescission after ACCEPTED/EXECUTED
• §7.3/Appendix A: Schema and trust-list references are name-based, not hash-pinned #16 — §7.3/App. A: Schema & trust-list references not hash-pinned

Medium

• §12: GDPR posture overstates "hash is not personal data" (EDPB Guidelines 02/2025) #17 — §12: GDPR "hash is not personal data" overstated (EDPB 02/2025)
• §5.3: Mandate verification discloses negotiation-sensitive caps to the counterparty #18 — §5.3: Mandate verification leaks negotiation-sensitive caps
• §7.3/§10: Penalties lack collateral — no performance bond for the performing party #19 — §7.3/§10: Penalties lack collateral (no performance bond)
• §6.5: Quantum security numbers for the anchor-hash requirement are imprecise #20 — §6.5: Quantum security numbers imprecise

Opportunities

• §10: Small-claims gap — the dispute path is uneconomical for exactly the micro-value cases the protocol targets #21 — §10: Small-claims process profile
• §7: Milestone / partial-performance support #22 — §7: Milestone / partial-performance support

Suggested order of attack

1. §6.1/§6.3: Signature input and hash coverage of proof[] are undefined #4, §7.4: N-party accepts conflict with the linear hash chain #5 — pure spec-text fixes, unblock any implementation work.
2. §6.2.1/§13: Uncontested enforcement requires contract-readable anchor state — missing DLT requirement #6, §6.5: PQC-gated status mutation is impractical to enforce on-chain as specified #7 — architectural decisions that shape the settlement-interface design; resolve before Phase-2 PoC.
3. §5.3: Mandate revocation enables TOCTOU/repudiation via a mutable HTTPS status list #8, §8.3/§9.2: VRF witness/arbiter selection is grindable #9, §9.4: No evidence window or ruling deadline — escrow can be stranded in CHALLENGED #10, §9.4/§6.2.1: Asserter bond is SHOULD-only — false assertions are free (and the two sections contradict each other) #12, §9.3/§7.4: Permissionless anchoring enables thread pollution and dispute-freeze griefing #13 — security/liveness; mostly additive normative rules.
4. Rest in severity order; §7.4/§10: No mutual-settlement fast path — every payment waits out the challenge window #14/§7: No amendment or mutual rescission after ACCEPTED/EXECUTED #15/§7: Milestone / partial-performance support #22 fit naturally together (they all touch mutual co-signed Objects and the basis: "mutual_settlement" directive).

[iret77]

All 19 review findings (plus the IOTA-profile issues #25–#30) are resolved on main as the v0.3 draft, via twelve PRs:

PR	Issues	Theme
#32	#4, #5	Signature input / hash coverage; chain vs. attachment linkage
#33	#6, #7, #26	Contract-readable dispute state; PQC status gate (hybrid: honest Core scoping + optimistic enforcement, MUST in the IOTA profile)
#34	#8, #27	Anchor-time mandate verification; status-list mechanism generalized (option a)
#35	#9, #10, #12, #13	Grind-resistant VRF seeds; evidence/ruling deadlines; mandatory asserter/challenger bonds; authorized-writer rule
#36	#14, #15, #22	Mutual settlement (settle), amend/rescind, milestones
#37	#16	Hash-pinned {id, hash} schema/trust-list/pool references
#38	#17, #20	GDPR pseudonymization posture (EDPB 02/2025); corrected quantum hash levels
#39	#25, #28, #29	IOTA profile: on-chain hash handling, realistic stablecoins, June-2026 refresh (Starfish, measured costs, randomness/ECVRF)
#40	#11, #19	Funded escrow gates EXECUTED; performance bond
#41	#21	Small-claims profile (minimal variant in v0.3 per maintainer decision; single-juror → v0.4)
#42	#18	SD-JWT predicate disclosure for mandates
#43	#30	IOTA Trust Framework positioning (candidate bindings, Phase-2 evaluation)

Every change is logged in SPEC.md Appendix C ("Changes from v0.2"). The spec is now v0.3-draft; no release/tag has been created (per RELEASING.md, releases only on explicit instruction). Closing this tracker.