feat(standards): pay the transaction fee in guarded and smart multisig auth by mmagician · Pull Request #3798 · 0xMiden/protocol

mmagician · 2026-09-03T09:55:00Z

Summary

Port of #3786 to next.

On next the guarded and smart multisig components decode their auth args but drop the conversion info and pay no fee. Both paths can authenticate below the account's spending quorum (guardian key rotation, or a procedure with a reduced per-procedure threshold), and the conversion rate is host-supplied, so paying the fee unbounded would let such an authorization move arbitrary value out of the vault as a fee note. This wires in the bounded flow from #3797 instead.

Notes

Differs from the rc version in following next's auth-args scheme: the conversion info comes from resolve_auth_args rather than load_conversion_info, so the auth args no longer double as the summary salt, and the summary is bound to the block number as on next.

The plain AuthMultisig component is bounded too, through the same multisig::pay_bounded_fee helper the guarded and smart components share, so this closes AuthMultisig: unbounded fee conversion lets approvers drain the vault via the fee note #3763 on next (the rc counterpart is feat(standards): bound the fee payment of the plain multisig component #3802). The MAST roots of all three multisig components change, so accounts built on them get new code commitments.

mmagician · 2026-09-03T10:35:24Z

+    # one slot beyond the approvers, for the guardian signature. It is unconditional because the
+    # rotation path verifies no guardian signature but scans every account procedure instead, which
+    # the slot also covers.


Suggested change

# one slot beyond the approvers, for the guardian signature. It is unconditional because the

# rotation path verifies no guardian signature but scans every account procedure instead, which

# the slot also covers.

# one slot beyond the approvers, for the guardian signature.

unnecessary impl. detail.
We should guard against such bloat in the future, add / modify the MASM skills (separate PR to do this)

mmagician · 2026-09-03T10:35:52Z

+    # settle the sponsorship obligation first, in pay_fee's order; the bound below guards the
+    # host-supplied rate, which the sponsorship amounts do not depend on


Suggested change

# settle the sponsorship obligation first, in pay_fee's order; the bound below guards the

# host-supplied rate, which the sponsorship amounts do not depend on

# settle the sponsorship obligation first

mmagician · 2026-09-03T10:37:45Z

 pub proc auth_tx_multisig_smart(auth_args: word)
+    # read the output-note count before the fee payment so the notes it creates can be counted
+    exec.tx::get_num_output_notes movdn.4


apply the same doc changes as suggested in the other file

mmagician · 2026-09-03T10:39:09Z

+    exec.multisig::get_initial_threshold_and_num_approvers drop
+    # => [num_of_approvers, CONVERSION_INFO, block_number, SALT, num_output_notes_before_fee]
+
+    exec.signature::estimate_multisig_authentication_cycles
+    # => [num_extra_cycles, CONVERSION_INFO, block_number, SALT, num_output_notes_before_fee]
+
+    exec.fee::estimate_fee
+    # => [fee_amount, CONVERSION_INFO, block_number, SALT, num_output_notes_before_fee]
+
+    # settle the sponsorship obligation first, in pay_fee's order; the bound below guards the
+    # host-supplied rate, which the sponsorship amounts do not depend on
+    exec.tx::get_fee_asset_id exec.fees::create_network_note_sponsorships drop
+    # => [fee_amount, CONVERSION_INFO, block_number, SALT, num_output_notes_before_fee]
+
+    dup movdn.5
+    # => [fee_amount, CONVERSION_INFO, fee_amount, block_number, SALT, num_output_notes_before_fee]
+
+    exec.fee::resolve_payment_info
+    # => [payment_faucet_id_suffix, payment_faucet_id_prefix, payment_amount, fee_amount,
+    #     block_number, SALT, num_output_notes_before_fee]
+
+    push.FEE_BOUND_DEN push.FEE_BOUND_NUM
+    # => [bound_num, bound_den, payment_faucet_id_suffix, payment_faucet_id_prefix, payment_amount,
+    #     fee_amount, block_number, SALT, num_output_notes_before_fee]
+
+    exec.fee::assert_fee_bound
+    # => [payment_faucet_id_suffix, payment_faucet_id_prefix, payment_amount, block_number, SALT,
+    #     num_output_notes_before_fee]
+
+    exec.fee::pay_estimated_fee
+    # => [block_number, SALT, num_output_notes_before_fee]
+
+    # the notes the fee payment created: the TX_FEE note and one FEE_SPONSORSHIP note per network
+    # output note. The procedure policies' note restrictions exclude them.
+    exec.tx::get_num_output_notes movup.6 sub
+    # => [num_own_output_notes, block_number, SALT]


isn't this logic duplicated in crates/miden-standards/asm/components/auth/guarded_multisig/guarded_multisig.masm? If we can re-use it, then let's pull it out to a common helper procedure

…g auth Port of #3786 to next. Both components run estimate_fee, the sponsorship payment, resolve_payment_info, assert_fee_bound (2/1, native fee asset) and pay_estimated_fee between multisig::resolve_auth_args and the summary, and pass the number of notes the payment created to guardian::verify_signature and multisig_smart::auth_tx, whose note restrictions exclude them. tx_policy::assert_no_output_notes reads the verified count itself. Adaptations to next: the conversion info comes from resolve_auth_args (MultisigAuthArgs) rather than load_conversion_info, so the auth args no longer double as the summary salt; multisig_smart::auth_tx takes (num_own_output_notes, block_number, SALT); the fee asset is read via tx::get_fee_asset_id; the sponsorship wrapper dropped in the fee-split port is spelled out at the call sites; pay_fee reads the fee asset once via an estimate_fee_for_asset helper. Tests build MultisigAuthArgs, share the multisig fixture and signing helpers, and drop the salt-binding assertions. Cost tables and bench-tx.json regenerated. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

…tisig auth Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

…sig auth Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

…g auth Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

…ig components Pull the fee-paying block of the guarded and smart multisig auth procedures into multisig::pay_bounded_fee, which takes the number of signers and the conversion info and returns the number of notes the payment created. The fee bound constants and the drain rationale live there once; the components keep only the signer count that differs. Written against fee::estimate_fee taking the fee asset ID, so the fee asset is read once for the estimate and the sponsorship notes. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

The estimate records every network note's sponsorship price in a local table, which the sponsorship payment reads back instead of pricing the notes through FPI a second time, mirroring fee::pay_fee. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

AuthMultisig now pays its fee through multisig::pay_bounded_fee, so the host-supplied conversion rate can move at most twice the computed fee out of the vault, in the native fee asset, closing the drain of #3763. The bound's rationale is reworded to cover every multisig component. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

Composing pay_fee from resolve_payment_info and pay_estimated_fee cost every plain fee payer 121 auth cycles and pushed the two-P2ID ECDSA consumption into the next padded-trace bracket. Only the bounded multisig payment composes the two procedures; pay_fee keeps its inline tail. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

The sponsorship price table now holds one fee asset value word per output note, so the local table grows to fees::SPONSORSHIP_PRICE_TABLE_NUM_ELEMENTS as in fee::pay_fee. The pointer plumbing is unchanged. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

mmagician commented Sep 3, 2026

View reviewed changes

mmagician mentioned this pull request Sep 3, 2026

chore: keep MASM inline comments to what the code does #3800

Merged

mmagician force-pushed the mmagician-claude/port-multisig-pays-fee branch from db6959d to cf7038d Compare September 3, 2026 11:02

mmagician mentioned this pull request Sep 3, 2026

feat(standards): bound the fee payment of the plain multisig component #3802

Open

mmagician force-pushed the mmagician-claude/port-multisig-pays-fee branch from cf7038d to 6f27e78 Compare September 3, 2026 13:46

mmagician force-pushed the mmagician-claude/port-multisig-pays-fee branch from 6f27e78 to 0611809 Compare September 3, 2026 14:32

claude added 12 commits September 3, 2026 15:01

docs(standards): shorten the guardian-slot comment in the guarded mul…

df2c1e1

…tisig auth Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

docs(standards): shorten the sponsorship comment in the guarded multi…

651f5ef

…sig auth Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

docs(standards): shorten the sponsorship comment in the smart multisi…

2e9331d

…g auth Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

docs(testing): drop the "sample" wording from the tx_policy test doc

97de4eb

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

chore: regenerate cost tables

bfe2ae7

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

chore: regenerate cost tables

78fe2df

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KkQzFtRjyDwQ7iVtRbsHsn

mmagician force-pushed the mmagician-claude/port-multisig-pays-fee branch from 0611809 to 78fe2df Compare September 3, 2026 15:10

mmagician requested a review from bobbinth September 3, 2026 15:12

Navigation Menu

Uh oh!

feat(standards): pay the transaction fee in guarded and smart multisig auth - #3798

feat(standards): pay the transaction fee in guarded and smart multisig auth#3798
mmagician wants to merge 12 commits into
mmagician-claude/port-fee-assert-boundfrom
mmagician-claude/port-multisig-pays-fee

mmagician commented Sep 3, 2026 •

edited

Loading

Uh oh!

mmagician Sep 3, 2026

Uh oh!

mmagician Sep 3, 2026

Uh oh!

mmagician Sep 3, 2026

Uh oh!

mmagician Sep 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		# settle the sponsorship obligation first, in pay_fee's order; the bound below guards the
		# host-supplied rate, which the sponsorship amounts do not depend on

Uh oh!

Conversation

mmagician commented Sep 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Notes

Uh oh!

mmagician Sep 3, 2026

Choose a reason for hiding this comment

Uh oh!

mmagician Sep 3, 2026

Choose a reason for hiding this comment

Uh oh!

mmagician Sep 3, 2026

Choose a reason for hiding this comment

Uh oh!

mmagician Sep 3, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

mmagician commented Sep 3, 2026 •

edited

Loading