Risk

The risks

Stream redirection

This is the sharpest one and it is structural, not hypothetical. Pons v2 lets the protocol propose a change to a token's creator fee recipient behind a three-day timelock. After three days anyone can carry it out; if nobody does within three more, it expires. The factory exposes pendingCreatorFeeRecipient(token) returning the proposed address and both deadlines.

So any protocol that derives value from a Pons fee stream can have that stream cut off with three days' notice, by Pons. EigenYield does not pretend otherwise: the registry watches the pending recipient and drops the pool from the basket the moment a proposal appears, which stops good money chasing a dying stream but does not refill what was already paid for it. Price your term accordingly, and read the term lengths in the ledger.

Silence

The vault is long activity. If Robinhood Chain goes quiet, harvests shrink toward zero and NAV goes flat. It does not fall, because there is no inventory to mark down, but flat is not free: you are holding an asset with an opportunity cost and no upside until people trade again. A factor bet on one chain's activity is exactly as concentrated as it sounds.

Contracts

The EigenYield vault and registry are unaudited. Pons v2, which they sit on, is also unaudited at the time of writing, with three engagements in progress and none closed. Locked liquidity removes the rug surface, not the bug surface.

Two more sharp edges worth knowing before you size a position. The Pons fee-escrow claim ABI has never been confirmed against deployed bytecode, so the registry calls it through a swappable encoder and accounts by balance delta rather than trusting a return value. And a seller contract that refuses its payout will revert the whole harvest batch; the mitigation is exclusion from the batch, which is a keeper action, not an automatic one.