These are the reports from Sail's AI security review by Octane (octane.security), an AI source-code security scanner. The review covered both the trusted core and the seven shared permission templates — not the core alone — across three successive security analyses (2026-06-24, 2026-06-26, 2026-06-29). The third and final analysis (2026-06-29) identified no critical- or high-severity findings; all reported vulnerabilities were resolved or acknowledged, and the remaining lower-severity warnings are documented, accepted by design, or out of scope.
What was reviewed. The trusted core (SailKernel, SailGovernance, the timelock, and the
core interfaces) together with the shared permission templates (SwapPermission,
SwapPermissionNoOracle, BorrowPermission, DepositPermission, WithdrawPermission,
TransferPermission, ApproveAndCallBatchPermission).
Review boundary. The reviews cover main through PR #79 (commit 8d1e122). Changes on
main after that point are limited to documentation and deployment metadata, not trusted-core
logic. Scope was defined by files and contracts; the reviewer did not state a line-count
figure.
Note: the
WithdrawPermissioncovered by these reviews was the earlier ERC-20-transfer gate. It has since been rewritten as the vault/lending-pool exit permission (ERC-4626withdraw/redeemand Aave v2/v3withdraw). The current vault-exit implementation is new code that postdates the review boundary above: it is not covered by these reports and has not been independently reviewed.
The reports are Octane's signed deliverables and are kept here as PDFs (they are not inline-embedded; open or download via the links below).
| Review | Date | Reviewer | Pages | Report |
|---|---|---|---|---|
| Third analysis (most recent) | 2026-06-29 | Octane | 153 | |
| Second analysis | 2026-06-26 | Octane | 175 | |
| First analysis | 2026-06-24 | Octane | 108 |
The most recent review (2026-06-29) reflects the current state of the reviewed code; earlier reviews are retained for transparency.
A security review is not a proof of correctness. It reduces risk in the reviewed code; it does not eliminate it, and it does not extend to code deployed by users. In particular, the correctness of any user-deployed permission or fee policy remains the author's responsibility. See the Known Limitations and Operator Responsibilities in the security model for the boundaries that hold by design.
See the canonical Security Policy at the repository root.