Neuron

Property	Value
Rust implementation	`NeuronPrecompile`
Solidity interface	`INeuron`
Address	`0x0000000000000000000000000000000000000804`
Status	Deployed

Registers neurons, publishes serving endpoints, and submits weights. Every function is payable.

Functions

burnedRegister(uint16,bytes32)
registerLimit(uint16,bytes32,uint64)
serveAxon(uint16,uint32,uint128,uint16,uint8,uint8,uint8,uint8)
serveAxonTls(uint16,uint32,uint128,uint16,uint8,uint8,uint8,uint8,bytes)
servePrometheus(uint16,uint32,uint128,uint16,uint8)
setWeights(uint16,uint16[],uint16[],uint64)
commitWeights(uint16,bytes32)
revealWeights(uint16,uint16[],uint16[],uint16[],uint64)

Added weight operations

Function	Source extrinsic
`setMechanismWeights`	`SubtensorModule.set_mechanism_weights`
`batchSetWeights`	`SubtensorModule.batch_set_weights`
`commitMechanismWeights`	`SubtensorModule.commit_mechanism_weights`
`batchCommitWeights`	`SubtensorModule.batch_commit_weights`
`revealMechanismWeights`	`SubtensorModule.reveal_mechanism_weights`
`commitCrv3MechanismWeights`	`SubtensorModule.commit_crv3_mechanism_weights`
`batchRevealWeights`	`SubtensorModule.batch_reveal_weights`
`commitTimelockedWeights`	`SubtensorModule.commit_timelocked_weights`
`commitTimelockedMechanismWeights`	`SubtensorModule.commit_timelocked_mechanism_weights`

Batch calls accept at most 16 outer items and at most 4,096 weight entries per inner array. Timelocked commits and registration work are bounded to 5,000 and 64 bytes respectively; none of these functions accepts SCALE-encoded runtime calls.

Added registration and key operations

Function	Source extrinsic
`register`	`SubtensorModule.register`
`rootRegister`	`SubtensorModule.root_register`
`swapHotkey`	`SubtensorModule.swap_hotkey`
`swapHotkeyV2`	`SubtensorModule.swap_hotkey_v2`
`setChildren`	`SubtensorModule.set_children`
`setIdentity`	`SubtensorModule.set_identity`
`tryAssociateHotkey`	`SubtensorModule.try_associate_hotkey`
`associateEvmKey`	`SubtensorModule.associate_evm_key`
`announceColdkeySwap`	`SubtensorModule.announce_coldkey_swap`
`executeAnnouncedColdkeySwap`	`SubtensorModule.swap_coldkey_announced`
`disputeColdkeySwap`	`SubtensorModule.dispute_coldkey_swap`
`clearColdkeySwapAnnouncement`	`SubtensorModule.clear_coldkey_swap_announcement`

Every added operation accepts a non-Root signed origin. The precompile dispatches the highest-level extrinsic as the mapped caller so runtime authorization remains in force. Root-only and deprecated compatibility calls are classified in the coverage audit.

Typed state views

getUid(uint16,bytes32)
isNetworkMember(bytes32,uint16)
getWeights(uint16,uint16)
getBonds(uint16,uint16)
getBlockAtRegistration(uint16,uint16)
getNeuronCertificate(uint16,bytes32)
getPrometheus(uint16,bytes32)
getChainIdentity(bytes32)
getSubnetIdentity(uint16)
getLoadedEmission(uint16)
getTransactionKeyLastBlock(bytes32,uint16,uint16)
getLegacyTransactionRateBlocks(bytes32)
getWeightCommitCount(uint16,bytes32)
getWeightCommit(uint16,bytes32,uint32)
getTimelockedWeightCommitCount(uint16,uint64)
getTimelockedWeightCommit(uint16,uint64,uint32)
getLegacyTimelockedWeightCommitCount(uint8,uint16,uint64)
getLegacyTimelockedWeightCommit(uint8,uint16,uint64,uint32)

The indexed commit readers expose stable metadata. Timelocked ciphertext is represented by its Keccak-256 hash and byte length instead of returning the runtime's encrypted payload type.

Proposed bulk runtime API

NeuronInfoRuntimeApi.get_neurons remains proposed. Its current result can grow with subnet size, so no Solidity selector is assigned in this change. Before implementation, it needs a bounded cursor-based or indexed interface with a stable typed result. Existing single-neuron and per-field views remain available through the deployed Neuron and Metagraph precompiles.

Source: neuron.sol