mev-pbs
Skill36 starsChanged 5 months agoMITView on GitHub
Detect MEV and proposer-builder-separation exposure — sandwichable swaps with no minOut, JIT liquidity, oracle-update frontrunning, backrunnable state, false reliance on private mempools, builder censorship, missing commit-reveal, and multi-block MEV post-PBS. Activate whenever code performs swaps/liquidations/auctions/redemptions whose ordering or price is observable in the public mempool before execution.
In this repository44 other files it gives its agents.
- access-controlSkill
- approval-issuesSkill
- caching-and-incrementalSkill
- centralization-riskSkill
- confidence-scoringSkill
- cosmwasmSkill
- cross-chain-messagingSkill
- cross-contract-stateSkill
- delegatecall-risksSkill
- diamond-eip2535Skill
- dos-vectorsSkill
- erc1271-contract-signaturesSkill
- erc4337-account-abstractionSkill
- erc4626-inflationSkill
- false-positive-feedback-loopSkill
- fee-on-transferSkill
- flash-loan-attacksSkill
- initializationSkill
- inline-assemblySkill
- integer-issuesSkill
- intents-erc7683Skill
- known-good-comparisonSkill
- liquidation-cascadeSkill
- mev-frontrunningSkill
- multi-pass-self-critiqueSkill
- oracle-manipulationSkill
- oracle-redundancySkill
- permit2-patternsSkill
- pragma-and-addressesSkill
- progress-and-streamingSkill
- reentrancySkill
- restaking-eigenlayerSkill
- selfdestruct-eip6780Skill
- signature-malleabilitySkill
- signature-replaySkill
- solana-anchorSkill
- storage-layoutSkill
- stylus-rustSkill
- token-compatibilitySkill
- tx-context-misuseSkill
- unchecked-callsSkill
- ve-lock-governanceSkill
- vyper-specificSkill
- zk-verifier-bugsSkill
The file
Files
MEV / PBS risk detection
When this applies
Trigger on any of:
- Swaps, mints, or redemptions priced from a spot source without slippage bounds
- Liquidations, auctions, or rebalances triggerable by anyone observing the mempool
- Logic assuming transactions land via a private orderflow (Flashbots Protect, MEV-Share)
- Reveal-then-act flows with no commit phase
- Reward/airdrop claims, NFT mints, or oracle pushes that are frontrun-profitable
- Time- or block-sensitive state an attacker can backrun
Detection patterns
Sandwichable swap, no minOut (HIGH)
function swap(uint256 amountIn) external {
uint256 out = router.swapExactTokensForTokens(amountIn, 0, path, ...);
// ^ minOut = 0
}Signal: amountOutMin == 0 (or a deadline of type(uint).max) lets a searcher frontrun to move price, then backrun, extracting the full slippage. The user-facing call must accept a caller-supplied minOut and deadline, never hardcode 0.
Oracle-update frontrunning (HIGH)
A position becomes profitable to liquidate / mint the instant a Chainlink push or updatePrice lands. Searchers backrun the oracle tx in the same block. If your protocol grants the liquidator a fixed bonus, the value leaks to MEV; if redemptions price off a freshly-updated feed, frontrun the update.
JIT liquidity (MEDIUM)
Concentrated-liquidity add-just-before / remove-just-after a large swap captures fees from passive LPs without bearing inventory risk. Flag fee mechanics that reward liquidity present only at swap time.
False private-mempool assumption (HIGH)
// comment: "safe because we submit via Flashbots, never public"Signal: security argument rests on transactions staying private. Private relays are best-effort, not guaranteed; post-PBS, builders may not include or may leak the bundle. On-chain logic must be safe even when public.
Missing commit-reveal (MEDIUM)
Auctions, randomness consumers, or fair-mints that reveal the winning input in the same tx are frontrunnable. No commit phase = bid sniping.
Multi-block / backrunnable state (MEDIUM)
Post-Merge, a proposer controlling consecutive slots can execute multi-block MEV (e.g. hold price across two blocks). State that's only manipulation-safe within a single block is not safe under multi-block control.
Severity rubric
| Pattern | Severity | Notes |
| minOut=0 / max deadline on user swap | High | Guaranteed sandwich loss |
| Oracle-update frontrun → liquidation/redemption | High | Value leaks to searchers |
| Security relies on private mempool | High | Assumption not enforceable on-chain |
| Missing commit-reveal on auction/mint | Medium | Bid sniping |
| JIT liquidity fee capture | Medium | LP value extraction |
| Single-block-only safety vs multi-block MEV | Medium | Proposer with consecutive slots |
Remediation patterns
- Caller-supplied
minOut+deadlineon every swap/redeem; reject 0 /type(uint).maxdefaults. - Commit-reveal for auctions, fair mints, and randomness consumption; or a sealed-bid scheme.
- Decouple action from oracle tick — use TWAPs / time-weighted settlement so a single update isn't instantly exploitable; cap per-block price movement.
- Don't rely on privacy for safety — treat the mempool as adversarial; if using MEV-Share/threshold encryption, still make on-chain logic safe when public.
- Multi-block-aware design — require state to be safe across consecutive proposer slots; consider per-epoch rate limits.
False-positive notes
- A swap whose
minOutis computed off-chain from a fresh quote and passed in is fine — don't flag the presence of a slippage parameter. - Functions gated to a trusted keeper/relayer with no public-mempool exposure narrow (not eliminate) the risk; note rather than escalate.
- Pure view/quote functions carry no MEV.
Related
- [[mev-frontrunning]] — overlapping general frontrunning patterns
- [[oracle-manipulation]] — spot-price moves are the substrate for sandwiches
- [[flash-loan-attacks]] — sandwich capital is often flash-borrowed
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

