Skip to Content
Developer FAQ

Developer FAQ

Common questions and pitfalls for integrators building on the v1.0.0 protocol. For user-side questions, see the user manual FAQ .

Reverts

WiringMismatch

A canonical-bundle gate failed. The calling contract resolved its dependencies, found that Furnace, MarketRouter, MineCore, VeClaimNFT, ClaimToken, or ShareholderRoyalties no longer agree on one canonical bundle, and reverted closed. This is intentional: split-brain deployments cannot silently mutate accounting.

Diagnose by reading the back-pointers on the live contracts and comparing them to deployments/<network>.json. If the manifest is stale relative to a redeploy, refresh it. If the live state has drifted from the manifest, the wiring needs governance attention. See Protocol Overview — Wiring safety model.

DelegatedEOA(address account) / DelegatedEOAOwner(address account)

The address carries an EIP-7702 designator (23-byte runtime starting with 0xEF0100). Owner seats, guardian seats, keeper allowlist seats, wiring-setter candidates, and delegation-session delegate seats all reject this case. Use a bare EOA or an ordinary contract instead. See Security, Guardian, Pausing — EIP-7702 designator rejection.

NotAContract()

A wiring setter was given an address with code.length == 0. Either it is a bare EOA (some setters that don’t care about 7702 specifically still require a contract) or the contract has not been deployed yet. MineCore and MaintenanceHub also use this error for the 7702 case instead of DelegatedEOA.

MinVeOutNotMet / MinClaimOutNotMet

A slippage floor tripped. The user-supplied minVeOut / minClaimOut exceeded what the Furnace would have produced at the live block. Either the quote moved after fetch, or the slippage tolerance is too tight. Re-quote via FurnaceQuoter and resubmit. The Furnace clamps minVeOut == 0 && veOut > 0 to 1 as a UX guard for the auto-compound and shareholder-compound paths; direct entry paths do not.

BonusTargetNotMet

A Make Offer / auto-furnace execution checked the live gross bonus against the saved target and the gate floored. The MarketRouter.bonusBpsVsPrincipalClaim view uses floor rounding via Math.mulDiv — the gate cannot spuriously reject a valid ratio. Wait for the bonus to recover or cancel the escrow.

TakeoversPaused / LockingPaused / TradingPaused

The guardian has paused the surface. See Security, Guardian, Pausing — Pause surfaces for the per-surface effect tables and what unwind paths remain available.

ListingCooldown

A market listing state change ran inside the 1-block cooldown. Wait one block and retry.

GenesisKingClaimNotCollected

MineCore.setTakeoversPaused(false) was called before the one-shot LaunchController.collectGenesisKingClaim(address) ran. This is a launch-window gate; on a fully launched deployment this revert should not appear.

Indexers

My subgraph misses enterWithToken rows

FurnaceEnter does not include tokenIn / amountIn in its payload — the values must be recovered from the transaction receipt. For top-level direct calls, calldata decoding works. For internal calls (Safe / smart account / helper / batch), the canonical recovery rule is the latest ERC20 Transfer log into the Furnace address (excluding CLAIM) strictly preceding the FurnaceEnter log. See Events and Indexing → “enterWithToken” bullet for the full rule. The shipped subgraph mapping (subgraph/src/mappings/furnace.ts, function deriveTokenEntryFromReceipt) is the reference implementation.

AutoMax extension stops accruing in my indexer

VeClaimNFT.setAutoMax(tokenId, enabled) rewrites lockEnd = block.timestamp + MAX_LOCK_DURATION on every flag-state change AND on the same-state enable refresh. The emitted AutoMaxSet(user, tokenId, autoMax) does not carry the new lockEnd. Indexers MUST refresh both the autoMax flag AND the persisted lockEnd (set to event.block.timestamp + MAX_LOCK_DURATION) on every event, regardless of flag-direction. See Events and Indexing → “AutoMax indexing invariant”.

The join key is (txHash, logIndex), not (txHash, user) — batched helper / maintenance transactions emit multiple sibling events that would otherwise collapse onto the same per-(txHash, user) row. The shipped subgraph requests receipt: true on the relevant handlers so it can walk the receipt and select the largest logIndex < execution.logIndex from the FurnaceEnter rows in the same transaction. See Events and Indexing → “TxFurnaceEnter join key”.

DelegationSessionUsed rows look corrupted

The same DelegationSessionUsed event selector is shared across every consuming contract (MineCore, Furnace, VeClaimNFT, LpStakingVault7D, ShareholderRoyalties, ClaimAllHelper, FurnaceGuardHelper). A (topic0) → ABI map collapses per-surface decode entries into one row. Scope the lookup by (contract, topic0). The shipped event-watcher composite-keys its decode index this way and admits the ClaimAllHelper address into its allowlist so harvest-claim delegation activity (P_SHAREHOLDER_CLAIM_FOR, P_LP_CLAIM_FOR) is observed alongside the rest.

Why does ShareholderAutoCompoundFailed fire without a paused state?

compoundForMany (the batch path) skips users instead of pausing them when the quote call fails. The single-user path pauses with a reason code. See Events and Indexing → “Baron auto-compound failures”.

SDK and tooling

Where is the SDK package?

The Agent SDK lives at agents/sdk/ inside the public protocol repo. It is not published as a standalone npm package in v1.0.0. See Agents and Automation — Install.

Which manifest do I read?

deployments/<network>.json is canonical. The .md mirror is rendered for humans. Always load the JSON in code and treat contracts.<Name>.address as the live address — implementation, proxyAdmin, and proxyAdminOwner are governance metadata for the proxy-backed quartet. See Runtime Proxy Upgrades — Canonical addresses.

How do I filter logs to skip pre-deployment blocks?

Every contract entry in the manifest carries a startBlock. Filter evt_block_number >= startBlock for that contract.

Where is the public subgraph?

There is no single public subgraph endpoint. Deploy the shipped subgraph against your own Graph node or self-hosted Graph Node. See Events and Indexing — Public subgraph endpoint.

Keepers and bots

My keeper’s payoutBoundViolation tripped — what happened?

The EOA balance delta did not equal −gasSpent after a transaction. The runtime guard treats this as a payout-routing leak (mining ETH or CLAIM rewards routed back to the keeper EOA instead of the intended recipient) and trips the circuit breaker. Verify reign / Collect / Furnace recipient configuration before resuming. See Maintenance and Bots.

Can I run a keeper that bypasses KEEPER_LIVE_RUN slippage guards?

No. KEEPER_ALLOW_UNSAFE_MIN_OUT and KEEPER_ALLOW_UNSAFE_MIN_VE_OUT are hard-forbidden when KEEPER_LIVE_RUN=1. This is enforced at config-load time and cannot be bypassed.

What’s the LaunchController.preflight() bitmask?

preflight() returns (uint256 statusBitmask, uint256 requiredSeedEth) mirrors every finalizeGenesis() precondition as packed bits 0..10 plus the would-be seed ETH. Use it for operator dry-run and monitoring instead of re-deriving the checks off-chain. See Genesis.

See also

Agent Plan FormatErrors Reference