Spec 466: cap staking scan fees at four hotkeys

[unarbos]

Cherry-pick of #3180 onto current main (spec 465 → 466).

Caps the staking-hotkey scan fee at four hotkeys. Full 256-hotkey execution weight stays for block admission. Native block budget goes 4s → 12s so a 65-item transfer_stake batch fits (PayTAO / Lium). EVM and scheduler budgets stay on the old 4s reference.

Type of Change

• Bug fix (non-breaking change which fixes an issue)

What this is

• Cap staking scan fees at four hotkeys #3180 commits e88d8fdb + 864992e16 rebased onto 7c732d8d (465).
• spec_version 466 so the train can move past live 465 on devnet/testnet.
• Required-CI hand fixes from Cap staking scan fees at four hotkeys #3180 run 35289224710:
  1. drop unused traits::Get in runtime/src/staking_fee.rs
  2. import SwapHandler in runtime/tests/alpha_only_coldkey_fees.rs so get_alpha_amount_for_tao resolves

What this is not

• No WeightInfo invention.
• No Add pallet-derivatives: longs and shorts on subnet alpha (spec 455) #3135.
• Baskets stay off on mainnet.
• Do not approve this train's mainnet env. Propose only. Inti should approve 466, not leftover 465 run 35267938998 or leftover 464 run 35262050154.

Checklist

• cargo fmt --check --all
• git diff --check
• The two required reds from Cap staking scan fees at four hotkeys #3180 fixed by hand
• 11 required gates (CI)

  

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
subtensor	Building	Preview	Sep 18, 2026 12:29am UTC

[unarbos]

@IntiTechnologies spec 466 Propose is the one to approve.

This is a fee cap for transfer_stake (and other stake-exit calls). After 464, the 256-hotkey scan bound made declared weight ~64× and broke client payouts (PayTAO / Lium and others). 466 charges the scan fee at four hotkeys. Full execution weight stays for block admission. Native block budget is 12s so a 65-item batch fits. EVM / scheduler stay on the old 4s budget.

Train: https://github.com/RaoFoundation/subtensor/actions/runs/35291554030
Merge: dfdfba159d675b532ac61d6ccea3745e34cf1bcc (#3181)
Live before this train: devnet 465, testnet 465, finney 464

Approve this 466 Propose, not leftover 465 on https://github.com/RaoFoundation/subtensor/actions/runs/35267938998 (cancelled; already deployed 465 everywhere except finney) and not leftover 464 Watch https://github.com/RaoFoundation/subtensor/actions/runs/35262050154.

BasketTradingEnabled stays off. Do not sudo-enable baskets. Do not merge #3135. Do not approve until I confirm Sudo-upgrade mainnet clone and test on this merge is green — I will report if that job goes red.

After you approve, signers:

btcli upgrade sign --url https://github.com/RaoFoundation/subtensor/releases/tag/v466 -w <your-wallet>

[github-actions]

🛡️ AI Review — Skeptic (security review)

VERDICT: SAFE

LOW contributor risk / baseline scrutiny: write-access maintainer with substantive merged history; no listed Gittensor association. Branch targets main; cherry-picked authorship is explained.

Reviewed all 10 changed files and traced fee discounts through dispatch weights, wrappers, and settlement. No dependency, CI, or review-policy changes.

Static analysis only; block timing remains unverified. git diff --check passed; git status --short was clean. Builds, tests, and formatting tools were not run under Skeptic restrictions.

Findings

No findings.

Conclusion

No malicious behavior or concrete security vulnerability found. Fee discounts preserve execution-weight accounting, while EVM and scheduler budgets retain their previous limits.

---

🔍 AI Review — Auditor (domain review)

VERDICT: 👍

Established contributor with write access and substantial merged history; Gittensor association UNKNOWN.

Reviewed all 10 changed files, fee settlement, wrapper traversal, RPC estimates, and added regression tests. Execution-weight accounting remains intact; EVM and scheduler budgets retain their previous limits. Spec version increases to 466.

This PR is the better candidate. Recommend closing #3180. Comparison with its final commit confirms the same implementation and tests plus the stated import fixes and spec updates. Other overlapping PRs address different features.

Validation: git diff --check passed; git status --short was clean. cargo fmt --check --all could not complete because rustup required writing to its read-only temporary directory. Python Ruff checks were skipped because uv and the existing SDK environment were unavailable. Builds and tests were left to CI; no files were modified.

Findings

No findings.

Conclusion

No substantive correctness issue found. The implementation matches the stated fee-cap policy and preserves full execution-weight accounting.

[github-actions]

🔄 AI review updated — Skeptic: SAFE Auditor: 👍

[unarbos]

@IntiTechnologies the 466 Propose from this merge (dfdfba1, 12s block budget) is void. Jacob dropped the block-budget change.

Approve the new fee-cap-only 466 instead: #3182 / train https://github.com/RaoFoundation/subtensor/actions/runs/35292102814 / merge cdffbe2f7ab0c37ab07884387bfbd6443dca178d.