Implement fast confirmation rule - #17122

terencechain · 2026-07-07T17:13:47Z

Implements the fast confirmation rule behind the --enable-fast-confirmation feature flag.

Adds a beacon-chain/confirmation package implementing the FCR algorithm: LMD-GHOST safety checks, FFG justification gates, epoch-boundary reconfirmation, and vote support aggregation
Runs once per slot at slot start after attestations are applied, under the forkchoice lock
Tracks the unrealized justified root per forkchoice node and adds read accessors for ancestry, voting source, and vote snapshots
Uses the confirmed root as the engine API safe block hash when the flag is enabled, falling back to unrealized justified otherwise
Wires up the fast_confirmation spec tests for altair through gloas (minimal preset), all 1190 vectors pass
Honors bls_setting: 2 in forkchoice spec test vectors, which ship unsigned blocks

syjn99

LGTM from multiple rounds of review - I believe the implementation is correct as per spec. We might optimize this more after merging this because the feature is gated by a flag and other changes in our production path apart from FCR looks safe.

Copilot

Pull request overview

Implements the Ethereum consensus “fast confirmation rule” (FCR) behind --enable-fast-confirmation, integrates it into beacon-node slot processing, exposes new forkchoice read APIs needed by the algorithm, and wires up minimal-preset spec tests.

Changes:

Add beacon-chain/confirmation package implementing FCR (safety thresholding + FFG gates + committee/vote aggregation) with unit tests.
Extend forkchoice with new read accessors (vote snapshots, ancestry queries, unrealized justification, slashed indices, confirmed payload hash) and track unrealized justified roots per node.
Add/enable FCR spec test runner + minimal spectests for Altair→Gloas, and wire FCR confirmed root into Engine API SafeBlockHash when enabled.

Reviewed changes

Copilot reviewed 54 out of 55 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
testing/spectest/shared/common/forkchoice/type.go	Extend forkchoice spectest check schema with FCR fields + meta bls_setting.
testing/spectest/shared/common/forkchoice/runner.go	Add fast-confirmation test runner path; refactor step handling helpers; honor `bls_setting: 2`.
testing/spectest/shared/common/forkchoice/builder.go	Add FCR-enabled builder variant; run FCR per-slot in spectest harness; adjust attestation disparity for FCR vectors.
testing/spectest/shared/common/forkchoice/BUILD.bazel	Add features dependency for FCR-enabled spectest builder.
testing/spectest/minimal/altair__fast_confirmation__fast_confirmation_test.go	Add minimal-preset Altair FCR spec test entrypoint.
testing/spectest/minimal/bellatrix__fast_confirmation__fast_confirmation_test.go	Add minimal-preset Bellatrix FCR spec test entrypoint.
testing/spectest/minimal/capella__fast_confirmation__fast_confirmation_test.go	Add minimal-preset Capella FCR spec test entrypoint.
testing/spectest/minimal/deneb__fast_confirmation__fast_confirmation_test.go	Add minimal-preset Deneb FCR spec test entrypoint.
testing/spectest/minimal/electra__fast_confirmation__fast_confirmation_test.go	Add minimal-preset Electra FCR spec test entrypoint.
testing/spectest/minimal/fulu__fast_confirmation__fast_confirmation_test.go	Add minimal-preset Fulu FCR spec test entrypoint.
testing/spectest/minimal/gloas__fast_confirmation__fast_confirmation_test.go	Add minimal-preset Gloas FCR spec test entrypoint.
testing/spectest/minimal/BUILD.bazel	Mark minimal spectest target large; include FCR spec test sources.
config/params/mainnet_config.go	Add `ConfirmationByzantineThreshold` forkchoice/FCR constant to mainnet config.
config/params/config.go	Add config field + YAML key for `CONFIRMATION_BYZANTINE_THRESHOLD`.
config/params/loader_test.go	Update placeholder spec field list for new config param.
config/features/flags.go	Add `--enable-fast-confirmation` feature flag.
config/features/config.go	Wire CLI flag into feature config.
changelog/terence_fcr.md	Add changelog entry for FCR feature.
beacon-chain/rpc/eth/config/handlers_test.go	Update spec endpoint test expectations for new config param exposure.
beacon-chain/forkchoice/types/types.go	Add `VoteData` type for exposing public vote snapshot fields to FCR.
beacon-chain/forkchoice/interfaces.go	Extend forkchoice getter interfaces with ancestry/vote snapshot/justification/slashed/confirmed-hash APIs.
beacon-chain/forkchoice/ro.go	Add locked RO delegations for newly-exposed forkchoice methods.
beacon-chain/forkchoice/ro_test.go	Extend RO locking tests/mocks for new RO forkchoice API surface.
beacon-chain/forkchoice/doubly-linked-tree/types.go	Track `unrealizedJustifiedRoot` per node for FCR justification queries.
beacon-chain/forkchoice/doubly-linked-tree/store.go	Store unrealized justified root on insert.
beacon-chain/forkchoice/doubly-linked-tree/store_test.go	Update tests for new store insert signature.
beacon-chain/forkchoice/doubly-linked-tree/unrealized_justification.go	Persist unrealized justified root during unrealized checkpoint updates.
beacon-chain/forkchoice/doubly-linked-tree/unrealized_justification_test.go	Update tests for new unrealized justified checkpoint setter.
beacon-chain/forkchoice/doubly-linked-tree/gloas.go	Fix vote node resolution when full-node entry is absent (pre-Gloas paths).
beacon-chain/forkchoice/doubly-linked-tree/forkchoice.go	Add new forkchoice read APIs required by FCR; adjust latest-message semantics; expose slashed/vote snapshot/confirmed payload hash.
beacon-chain/forkchoice/doubly-linked-tree/forkchoice_test.go	Add tests for new forkchoice helper APIs and updated vote semantics.
beacon-chain/core/transition/trailing_slot_state_cache.go	Add `ClearNextSlotCache` helper for isolating spectest runs.
beacon-chain/confirmation/types.go	Add shared types for FCR (equivocation scorer + balance info container).
beacon-chain/confirmation/interfaces.go	Define forkchoice/committee/balance accessor interfaces for FCR integration.
beacon-chain/confirmation/helpers.go	Implement committee weight estimation + adversarial/proposer scoring helpers.
beacon-chain/confirmation/helpers_test.go	Unit tests for committee weight/adversarial/proposer helper logic.
beacon-chain/confirmation/support.go	Implement vote support aggregation and equivocation scoring helpers for FCR queries.
beacon-chain/confirmation/support_test.go	Unit tests for support aggregation, equivocation, and table rebuild behavior.
beacon-chain/confirmation/safety.go	Implement LMD-GHOST safety thresholding + `is_one_confirmed`.
beacon-chain/confirmation/safety_test.go	Unit tests covering safety threshold, epoch crossing, optimistic blocks, and slot-0 behavior.
beacon-chain/confirmation/ffg.go	Implement FFG target support estimation + justification gate helpers.
beacon-chain/confirmation/ffg_test.go	Unit tests for FFG target scoring + honest-support computations.
beacon-chain/confirmation/confirmation.go	Implement full FCR store + per-slot update/advance/revert algorithm.
beacon-chain/confirmation/confirmation_test.go	Unit tests for key FCR store rotation + phase behaviors.
beacon-chain/confirmation/BUILD.bazel	Add Bazel targets for FCR library and tests.
beacon-chain/blockchain/service.go	Initialize FCR when flag enabled; compute engine SafeBlockHash from confirmed root when available.
beacon-chain/blockchain/receive_attestation.go	Run FCR at slot start after applying attestations.
beacon-chain/blockchain/receive_execution_payload_envelope.go	Use new safe hash selection (FCR confirmed root when enabled).
beacon-chain/blockchain/execution_engine.go	Use new safe hash selection (FCR confirmed root when enabled).
beacon-chain/blockchain/receive_block.go	Allow disabling block signature verification for spectests with unsigned blocks; add test helper method.
beacon-chain/blockchain/chain_info_forkchoice.go	Expose `Service.FCR()` accessor for spectest builder/harness.
beacon-chain/blockchain/fcr_accessors.go	Implement committee/balance accessors bridging Service state to FCR interfaces.
beacon-chain/blockchain/fcr_accessors_test.go	Add test ensuring checkpoint balance source handles skipped epoch boundary correctly.
beacon-chain/blockchain/execution_engine_test.go	Minor formatting fix in test setup call.
beacon-chain/blockchain/BUILD.bazel	Wire new FCR accessors + confirmation dep into blockchain build and tests.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

+			if uint64(v) >= uint64(len(offs)) {
+				grown := make([]uint8, v+1)
+				copy(grown, offs)
+				for i := len(offs); i < len(grown); i++ {
+					grown[i] = noAssignment
+				}
+				offs = grown
+			}


 	currentSlot := uint64(tick) / params.BeaconConfig().SecondsPerSlot
 	for lastSlot < currentSlot {
 		lastSlot++
 		bb.service.SetForkChoiceGenesisTime(time.Now().Add(-1 * time.Duration(params.BeaconConfig().SecondsPerSlot*lastSlot) * time.Second))


+		}
 	}
 	if tick > int64(params.BeaconConfig().SecondsPerSlot*lastSlot) {
 		bb.service.SetForkChoiceGenesisTime(time.Now().Add(-1 * time.Duration(tick) * time.Second))


Copilot

Pull request overview

Copilot reviewed 59 out of 60 changed files in this pull request and generated 3 comments.

+	confirmedRoot := f.confirmedRoot
+	currentEpoch := slots.ToEpoch(currentSlot)
+	// The slot start snapshot keeps every phase on one head even if forkchoice moves mid run.
+	head := f.currentSlotHead
+


+	if slots.IsEpochStart(currentSlot + 1) {
+		ujc := f.fc.UnrealizedJustifiedCheckpoint()
+		f.previousEpochGreatestUnrealizedCheckpoint = *ujc
+	}


potuz · 2026-09-02T20:26:12Z

+// DisableBlockSignatureVerificationForTesting supports spectest vectors generated with bls_setting 2, their blocks are unsigned.
+func (s *Service) DisableBlockSignatureVerificationForTesting() {
+	s.skipBlockSignaturesForTesting = true
+}


Ohhh came here to write essentially the same :)

@potuz what do you think of this way:

Add fake BLS backend for bls_setting: 2 spectest cases #17281

so we don't have to add skipBlockSignaturesForTesting field like this PR, and skip only for testing purpose via build flag.

# Conflicts: # beacon-chain/forkchoice/ro_test.go # beacon-chain/rpc/eth/config/handlers_test.go # testing/spectest/minimal/BUILD.bazel

potuz

Another quick pass, everything seems fine so far, a couple more of error handling that may result in a mistake, mostly nits this time.

potuz · 2026-09-03T13:12:37Z

+	}
+	balances := make([]uint64, st.NumValidators())
+	epoch := coreTime.CurrentEpoch(st)
+	for idx, val := range st.ValidatorsReadOnlySeq() {


Here we do a whole validator sweep and it seems to me that the balances are not cached on PulledUpHeadState. BalanceInfoByCheckpoint does cache, should we cache here at least on the pair (head, epoch)?

potuz · 2026-09-03T13:14:19Z

+		return nil, 0, err
+	}
+	if a.byCheckpoint == nil {
+		a.byCheckpoint = make(map[forkchoicetypes.Checkpoint]*confirmation.FFGStateInfo)


This map is written under no lock. I think this is fine as FCR is single threaded, but then my comment on multiple calls to getFFG in the same slot can actually be a race if we every hit that path.

potuz · 2026-09-03T13:34:30Z

+	if s.fcr != nil {
+		root := s.fcr.ConfirmedRoot()
+		if root != ([32]byte{}) {
+			return s.cfg.ForkChoiceStore.ConfirmedPayloadBlockHash(root)


Can't the payloadBlockHash here return zero as well? is this a problem? shouldn't in this case return the uj one?

…eshold

…dates

…th a mutex

…missing

potuz

This time I went over locks, there is no race now because there's a single thread that reads/writes I believe, but still there are some read locks being used to write and some fields read without any lock.

Those perhaps are not that important, but the O(n) under a forkchoice lock seems like a bad idea and can be avoided I think

potuz · 2026-09-03T14:41:12Z

+	f.fc.RLock()
+	defer f.fc.RUnlock()


This lock is taking over forkchoice and can stall block import. It does a full validator's sweep and it seems to me that the expensive part can be taken out of the lock, will mark in the relevant places.

potuz · 2026-09-03T14:41:52Z

+	equivocating := f.fc.SlashedIndices()
+	f.votesBuf = f.fc.VoteSnapshot(f.votesBuf[:0])
+	votes := f.votesBuf


Here we snapshot the votes (still taking about the fc.Lock)

potuz · 2026-09-03T14:45:32Z

+	equivocating := f.fc.SlashedIndices()
+	f.votesBuf = f.fc.VoteSnapshot(f.votesBuf[:0])
+	votes := f.votesBuf
+	f.support.Build(votes, balances, f.tables, equivocating, f.fc)


Then we call this function that makes an O(n) loop on validators, but all the arguments are already copied from forkchoice, so this loop can be taken out of the lock itself, is there a race by doing so?

Yes there is a race, the vote loop itself reads nothing from forkchoice but Build also walks the ancestry with ParentRoot to push support up, that part races with inserts. Split in b02664e, the O(n) aggregation runs off the lock and only the ancestor walk stays under it.

potuz · 2026-09-03T14:52:13Z

+	}
+
+	// Checkpoint state loads can replay from disk, keep them off the forkchoice lock.
+	balances, totalActiveBalance, err := f.balances.BalanceInfoByCheckpoint(ctx, f.currentEpochObservedJustifiedCheckpoint)


This is a read without a lock 1/3

potuz · 2026-09-03T14:52:33Z

+	var prevBalances []uint64
+	prevTotalActive := uint64(0)
+	if slots.IsEpochStart(currentSlot) {
+		prevBalances, prevTotalActive, err = f.balances.BalanceInfoByCheckpoint(ctx, f.previousEpochObservedJustifiedCheckpoint)


This is a read without a lock 2/3

potuz · 2026-09-03T14:52:49Z

+
+	// honest() needs the pulled up head state only on boundary, catch-up, or revert runs.
+	getFFG := sync.OnceValue(func() *FFGStateInfo {
+		ffg, err := f.balances.PulledUpHeadState(ctx, f.currentSlotHead)


This is a read without a lock 3/3

potuz · 2026-09-03T14:55:08Z

+		fastConfirmationDuration.WithLabelValues(path).Observe(float64(time.Since(start).Milliseconds()))
+	}(time.Now())
+
+	f.fc.RLock()


This takes a read lock, but writes some

previousSlotHead [fieldparams.RootLength]byte currentSlotHead [fieldparams.RootLength]byte currentEpochObservedJustifiedCheckpoint forkchoicetypes.Checkpoint previousEpochObservedJustifiedCheckpoint forkchoicetypes.Checkpoint previousEpochGreatestUnrealizedCheckpoint forkchoicetypes.Checkpoint

That are read unguarded in a few places marked below.

potuz

Pointing that we seem to be not even reading the confirmed head from the forkchoice tests vectors.

potuz · 2026-09-03T15:39:21Z

+	CurrentSlotHead                           *string    `json:"current_slot_head"`
+	PreviousEpochObservedJustifiedCheckpoint  *EpochRoot `json:"previous_epoch_observed_justified_checkpoint"`
+	CurrentEpochObservedJustifiedCheckpoint   *EpochRoot `json:"current_epoch_observed_justified_checkpoint"`
+	PreviousEpochGreatestUnrealizedCheckpoint *EpochRoot `json:"previous_epoch_greatest_unrealized_checkpoint"`


it seems to me we are not checking the fast confirmed head in tests

confirmed_root was already checked in the builder, but you're right there was a gap, we were silently dropping safe_execution_block_hash from the checks. Added in 206559f

potuz

Last round, still have the dominant lock of O(n) and the in-state computation of unrealized justification to cover, but other than that it seems good to go.

potuz · 2026-09-03T18:04:32Z

+		}
+		for _, t := range s.tables {
+			if sl, ok := t.assignedSlot(idx); ok && sl >= startSlot && sl <= endSlot {
+				total += s.balances[idx]


I think here we are counting equivocators that are onchain with balance zero and the spec would count this balance 1/2

potuz · 2026-09-03T18:05:43Z

+	balances := make([]uint64, st.NumValidators())
+	epoch := coreTime.CurrentEpoch(st)
+	for idx, val := range st.ValidatorsReadOnlySeq() {
+		if helpers.IsActiveValidatorUsingTrie(val, epoch) && !val.Slashed() {


Here we do not count slashed validators in the balance 2/2.

I believe in the previous message (marked 1/2) we needed to count the slashed balance. This is the whole mechanism to recover fast confirmation even in the presence of slashing.

Fixed in 059783f. Checked all the other consumers, attestation score, block support between slots, and current target score all want unslashed per spec

potuz

I did my best on reviewing this one thoroughly, it's such a massive change that probably more eyes would be good anyway. I did find a series of bugs, and one is remaining which is the issue of the unrealized justification not being computed before slot 21, I think it's safe anyway and fine to ship right now, but we should open an issue to fix that

…s#17281) **What type of PR is this?** > Other: Spectest harness **What does this PR do? Why is it needed?** - OffchainLabs#17122 The PR above says "Honors `bls_setting: 2` in forkchoice spec test vectors, which ship unsigned blocks". `bls_setting` is not yet widely used, but in the FCR context, ALL FCR tests are based on `bls_setting: 2` for faster test duration. OffchainLabs#17122 solves this problem ad-hoc like adding `skipBlockSignaturesForTesting` field and skips the signature verification like: ```go if s.skipBlockSignaturesForTesting { _, postState, err = transition.ExecuteStateTransitionNoVerifyAnySig(ctx, preState, signed) } else { postState, err = transition.ExecuteStateTransition(ctx, preState, signed) } ``` I think this can be much improved by enforcing so-called "fake" BLS backend that returns "yes" for every signature verification request. This compile-time approach has several benefits: - We don't have to touch the production path like above. - Some tests like `test_is_one_confirmed_fails_recently_activated_validator_voting_in_empty_slot` assumes a deposit process. We should also add some field like `skipDepositSignatures` in order to include the new deposit request (with stub signature). **Which issue(s) does this PR fix?** N/A **Other notes for review** Verified with the latest spectest fixtures + merge this work with `fcr` branch and test. Here's a matrix for every cases regarding `meta.yaml` and `bls_seting`: | Case's `meta.yaml` | Meaning in the spec | `bls=real` (default) | `bls=fake` | | --- | --- | --- | --- | | no `meta.yaml` | no metadata at all | ✅ run | ✅ run | | no `bls_setting` key (only `blocks_count`, etc.) | unspecified = same as `0` | ✅ run | ✅ run | | `bls_setting: 0` | BLS optional — valid either way | ✅ run | ✅ run | | `bls_setting: 1` | BLS required — only valid with BLS on | ✅ run | ⏭️ skip | | `bls_setting: 2` | BLS ignored — only valid with BLS off | ⏭️ skip | ✅ run | | any other value (e.g. `3`) | unknown | ❌ hard fail | ❌ hard fail | | `realOnlyCases`, 3 entries (`utils.go:78`) | no metadata, but needs real verification | ✅ run | ⏭️ skip | **Acknowledgements** - [x] I have read [CONTRIBUTING.md](https://github.com/prysmaticlabs/prysm/blob/develop/CONTRIBUTING.md). - [x] I have included a uniquely named [changelog fragment file](https://github.com/prysmaticlabs/prysm/blob/develop/CONTRIBUTING.md#maintaining-changelogmd). - [x] I have added a description with sufficient context for reviewers to understand this PR. - [x] I have tested that my changes work as expected and I added a testing plan to the PR description (if applicable).

terencechain force-pushed the fcr branch 4 times, most recently from 3e840d2 to 4e64fd7 Compare July 7, 2026 23:29

terencechain changed the title ~~Implement fast confirmation rule (FCR)~~ Implement fast confirmation rule Jul 7, 2026

syjn99 reviewed Jul 9, 2026

View reviewed changes

Implement FCR

2ebb970

terencechain force-pushed the fcr branch from 3e4f050 to 2ebb970 Compare July 10, 2026 16:46

syjn99 self-requested a review July 13, 2026 05:16

syjn99 reviewed Jul 15, 2026

View reviewed changes

Comment thread beacon-chain/blockchain/fcr_accessors.go Outdated

Comment thread beacon-chain/confirmation/support.go Outdated

Advance checkpoint state to epoch start for FCR balance source

a869dda

syjn99 mentioned this pull request Jul 19, 2026

Fast Confirmation Rule #15399

Open

syjn99 self-requested a review July 21, 2026 06:57

syjn99 previously approved these changes Jul 21, 2026

View reviewed changes

FCR: Performance improvements regarding SupportMap (#17179)

854cab3

Copilot AI review requested due to automatic review settings July 21, 2026 17:53

terencechain dismissed syjn99’s stale review via 854cab3 July 21, 2026 17:53

Copilot started reviewing on behalf of terencechain July 21, 2026 17:54 View session

Copilot AI reviewed Jul 21, 2026

View reviewed changes

syjn99 previously approved these changes Jul 22, 2026

View reviewed changes

FCR: Add fast_confirmation SSE topic (#17220)

5fc7c60

Copilot AI review requested due to automatic review settings July 22, 2026 22:07

terencechain dismissed syjn99’s stale review via 5fc7c60 July 22, 2026 22:07

Copilot started reviewing on behalf of terencechain July 22, 2026 22:07 View session

Copilot AI reviewed Jul 22, 2026

View reviewed changes

This was referenced Jul 30, 2026

[DRAFT] Compute real deposit signatures when BLS verification is disabled syjn99/consensus-specs#1

Draft

Add fake BLS backend for bls_setting: 2 spectest cases #17281

Merged

samcm mentioned this pull request Aug 21, 2026

Add Prysm FCR simulator engine ethpandaops/fcr-simulator#14

Merged

syjn99 and others added 2 commits August 25, 2026 06:15

FCR: Add metrics (#17218)

abf4f6e

Merge remote-tracking branch 'origin/develop' into fcr

94c2ace

# Conflicts: # beacon-chain/forkchoice/ro_test.go # beacon-chain/rpc/eth/config/handlers_test.go # testing/spectest/minimal/BUILD.bazel

potuz reviewed Sep 3, 2026

View reviewed changes

Propagate safety threshold errors instead of confirming on a zero thr…

701dce1

…eshold

terencechain mentioned this pull request Sep 3, 2026

Expose a payload-content aware IsOptimistic for gloas #17451

Open

terencechain added 3 commits September 3, 2026 07:52

Bail on getCurrentTarget error and skip repeated per-slot variable up…

4eb4e20

…dates

Cache the pulled up head state balances and guard the FCR accessor wi…

2d4e3dd

…th a mutex

Fall back to unrealized justified when the confirmed payload hash is …

31905b5

…missing

terencechain force-pushed the fcr branch from cee5fc1 to 31905b5 Compare September 3, 2026 14:54

potuz reviewed Sep 3, 2026

View reviewed changes

terencechain added 2 commits September 3, 2026 09:03

Move the O(n) support aggregation off the forkchoice lock

b02664e

Check safe_execution_block_hash in fast confirmation spec tests

206559f

potuz reviewed Sep 3, 2026

View reviewed changes

Count slashed validators in the equivocation score

059783f

potuz previously approved these changes Sep 4, 2026

View reviewed changes

terencechain added this pull request to the merge queue Sep 4, 2026

github-merge-queue Bot removed this pull request from the merge queue due to failed status checks Sep 5, 2026

Merge branch 'develop' into fcr

4a1403a

terencechain enabled auto-merge September 5, 2026 02:34

terencechain added 2 commits September 5, 2026 07:11

Merge branch 'develop' of github.com:OffchainLabs/prysm into fcr

18986b7

Rotate FCR slot head variables on every on_fast_confirmation call

e6b6ddd

terencechain dismissed potuz’s stale review via e6b6ddd September 5, 2026 14:12

Update bls mock for spec tests touching deposits

055ad52

syjn99 approved these changes Sep 7, 2026

View reviewed changes

terencechain added this pull request to the merge queue Sep 7, 2026

Merged via the queue into develop with commit e038e27 Sep 7, 2026
36 of 37 checks passed

terencechain deleted the fcr branch September 7, 2026 05:59

This was referenced Sep 7, 2026

Remove skipBlockSignaturesForTesting bypass #17461

Open

FCR spectest for v1.7.0-beta.0 #17467

Open

Uh oh!

Conversation

terencechain commented Jul 7, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

syjn99 left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

potuz left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

potuz left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

potuz left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

terencechain Sep 3, 2026 •

edited

Loading