On-chain cover promises protection against hacks and failures. Whether it delivers depends on wording, capital, claims processes and governance. We dissect one provider at a time: what is covered, who decides, and which risks the buyer keeps. Facts, not advice.
Underwriters we score on the Raccoon rubric: does the policy pay for the risk it advertises?
For: DeFi users who want to cover their own positions (and teams via the builder products)
Transparent, battle-tested, well capitalised relative to active cover. The structural weakness is legal: payout is a discretionary member decision, not an enforceable claim.
For: Protocol teams combining an audit with a coverage commitment (payout goes to the TEAM)
A narrow but transparently framed product. The dispute process with UMA escalation is a distinctive design and the Euler payout is a documented track record. On the other side, coverage is capped at $500k, terms are private per engagement, and Sherlock states that neither payment nor availability of funds is guaranteed. The Raccoon Score rates Shield on the builder rubric, for its actual buyer, the protocol team; retail users have no claim under this product at all. The low score reflects the private terms, the $500k cap, the non-guaranteed funds and that the payout is not committed to reach the users, and is not a verdict on Sherlock as an auditor, which is top-tier.
Aggregators and intermediaries. Scored on a different rubric: the duties of a cover broker (market access, advice, best execution, placement, claims advocacy, care, underwriter vetting), not the cover-payout score.
One scale, two rubrics: the Raccoon Score, 0 to 100 points across seven categories. Retail covers (you buy for yourself) are scored on the retail rubric; team covers (a protocol team buys, e.g. Native Protocol Cover or Sherlock Shield) get their own builder rubric, because a team asks different questions. Facts, provider statements and our own judgements are kept apart, every material claim is sourced, and anything we cannot verify is labelled as such instead of being estimated. Each analysis is a snapshot with a visible date.
Three separate lenses, never compared against each other. The retail rubric weighs the technical security of the covered protocol; the builder rubric replaces that with whether a payout reaches the protocol users; the broker rubric scores an intermediary on the duties of a cover broker (access, advice, execution, placement, claims advocacy, care, underwriter vetting), not a policy. The ranking lists them apart.
The score is an opinion based on public sources, not a probability of payout and not a guarantee. No legal, investment or insurance advice.