docs(tbtc/signer): record 2026-06-12 architecture decisions in gates doc - #4043
Conversation
- external audit covering frost-core 3.x + frost-secp256k1-tr is a HARD GATE for the ECDSA-retirement phases (resolves the Gate 1 fork recorded earlier today) - sidecar signer process chosen over in-process cgo as the target architecture; dlopen bridge stays transitional; unblocks #4007 scoping - script-tree commitment vs timelocked recovery leaf: explicitly open, no assumption may be baked in yet - proof-carrying blame deferred until production WITH a binding retention condition: keep enough signed bytes at detection points to diagnose targeted equivocation - t-of-included finalize scheduled as the first Phase 7 item: the transitional flow binds shares to the full included set's commitment list at StartSignRound (finalize enforces contributions == included set), so first-t-responsive requires the interactive two-round exchange that is Phase 7 itself Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
…etion Decision 6 (2026-06-12, MacLane): the transitional deterministic-nonce path is dev/staging-only behind the production gate, and its nonce safety rests on RoundNonceBinding transcript completeness - the F1 finding showed one missing field is a key-extraction-class bug that an experienced review missed. No production benefit justifies carrying that invariant indefinitely. - deletion trigger: interactive production path validated end to end; then the transitional StartSignRound/FinalizeSignRound deterministic flow and the nonce-binding machinery are removed - until then the transitional flow is FROZEN: no new transcript inputs (each must extend RoundNonceBinding; omission recreates F1) - nonce.rs carries the freeze marker at the point of hazard - item 6 interaction recorded: the Phase 7 interactive session flow is designed t-of-included-native from the start; no retrofit of the transitional finalize contract Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ion log Codex and Gemini both flagged (P1): the item-7 deferral parenthetical asserted evidence retention "is implemented in the Go RFC-21 layer", but the base branch only detects a conflict and drops the envelope - the retention logic lives in the unmerged PR #4044. Because item 7's deferral is conditioned on retention being present, that false claim created a false sense of diagnosability. Reworded: the deferral is now explicitly contingent on retention landing; retention is attributed to PR #4044 (scaffold branch); and the base layer's drop-the-envelope behavior until that merges is stated plainly, so the deferral does not read as already in force. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Codex + Gemini P1 (both identical) — valid, fixed in Reworded so the decision log is honest about sequencing: the deferral is now explicitly contingent on retention landing; retention is attributed to PR #4044 (scaffold branch); and the base layer's drop-the-envelope behavior until that merges is stated plainly, so the deferral does not read as already in force. Merge ordering is now self-documenting — if this doc lands before #4044, it correctly describes the gap rather than papering over it. The reviewers' other verifications (Phase 7 / t-of-included rationale, the nonce-path freeze, audit-gate phrasing, recovery-leaf openness) came back clean — no changes there. |
…oints (#4044) Implements the binding retention condition from today's decision log (PR #4043): proof-carrying blame (follow-up item 7) is deferred until production, **provided** telemetry/logging retain enough signed bytes to diagnose whether targeted equivocation is occurring — otherwise the revisit condition lacks data. ## What this adds `EquivocationEvidence` events carrying the **exact signed snapshot envelopes** (wire bytes verbatim — the #4040 format makes these available at every detection point) for the three detections that exist today: - `snapshot_conflict` — a sender re-submits a *different* signed snapshot for the same attempt to the coordinator. Both envelopes are retained; two operator-signed bodies from the same sender for the same attempt are self-incriminating, which is exactly the substrate item 7's wire format will formalize. - `own_snapshot_mutated_in_bundle` — a bundle carries this member's snapshot with a signature that differs from what it submitted (both envelopes retained). - `own_snapshot_missing_from_bundle` — censorship detection (self envelope retained). Each event is logged in full (these are rare, and the bytes are the diagnosis) and forwarded to a process-wide observer hook following the repo's existing single-observer telemetry pattern, so hosts can persist evidence into their telemetry stack. Emission is purely additive on the existing error paths — encode failures degrade to nil fields with a log line, never perturbing the protocol path. ## Deliberately out of scope Cross-member comparison (a receiver checking a bundle's snapshot for sender X against X's direct broadcast) — that's item 7 proper. These are the detection points that exist today, instrumented so the production deferral is honest. Tests pin byte-exact envelope retention for all three kinds, and that an idempotent identical re-submission emits nothing. `go build ./...`, vet, gofmt clean; full frost suite green. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
ee80aef
into
extraction/frost-signer-mirror-2026-05-26
1 participant
Records MacLane's decisions on the follow-up checklist's open questions: audit = hard gate for ECDSA retirement; sidecar over in-process cgo; script-tree vs timelock explicitly open; item 7 deferred with a binding evidence-retention condition; item 6 scheduled as the first Phase 7 item (with the architectural reason: transitional shares are bound to the full included set at StartSignRound, so first-t-responsive requires the interactive exchange that is Phase 7 itself).
🤖 Generated with Claude Code