Breakage analysis per dependency
BENQI's assessment for RD-F-052 — scored yellow on the v1.7.0 rubric. The evidence below is the curator's reasoning for this score.
Evidence summary #
Per-dependency breakage analysis (transcribed from this agent's 03-oracle-deps.md row, line 83). (1) Chainlink feed stale/halted: BENQI core lending markets continue serving last-known price (no staleness reversion in BenqiChainlinkOracle.getUnderlyingPrice()), creating under/over-collateralisation risk during extended price moves; worst case an Avalanche Chainlink network halt allows positions to become undercollateralised without triggering liquidations. (2) BenqiChainlinkOracle admin compromise: malicious setFeed() redirects asset pricing → collateral drain via inflated borrow limits (no timelock on oracle admin). (3) Chaos Labs Edge Oracle failure: Isolated Markets fall back to Chainlink secondary per the dual-oracle design (mitigated). (4) Avalanche P-Chain halt: sAVAX staking pauses but exchange rate is stable (cannot decrease — Certora-proven), LST holders not immediately harmed. Net: single non-redundant external dependency (Chainlink) on the core lending markets with no on-cha
Sources #
- AuditDedaub BENQI Ignite auditDedaub BENQI Ignite audit (Mar 2023) — M1 staleness-pattern finding contextretrieved 2026-05-16
- BenqiChainlinkOracle sourceBenqiChainlinkOracle verified source (Snowscan) — getUnderlyingPrice has no updatedAt staleness guardretrieved 2026-05-16
- Certora BENQI sAVAX reportCertora sAVAX formal verification — exchange-rate-monotonicity proof (P-Chain-halt branch)retrieved 2026-05-16
Methodology #
Produce a short per-dependency text describing which protocol functions halt or degrade and impact severity if each declared dependency fails.
See the full factor methodology and distribution across all protocols →