Content
50%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The content is well-structured and reasonably actionable but relies on pseudocode in places, lacks validation feedback loops for risky operations, and is monolithic with no progressive disclosure to bundle files.
Suggestions
Replace the comment-only Solidity blocks with executable contract snippets, or explicitly justify the pseudocode as illustrative.
Add an explicit validation/simulation checkpoint (e.g. simulate the bundle on a fork and confirm net profit before submitting) before the send step.
Move the tooling list and code patterns into reference files (e.g. references/patterns.md) and link to them from SKILL.md to enable progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly code and lean concrete guidance, but it opens by restating the sandwich concept ("buy the same token before the victim...") and includes prose such as "Sandwich attacks are public mempool predation, generally legal but ethically grey" that Claude already knows, so it could be tightened. | 2 / 3 |
Actionability | Real Python/JS snippets are present, but the Solidity blocks are comment-only pseudocode ("// Bundle: 3 tx in same block") and the JS example references undefined variables (attackerTx1, rawVictimTx), leaving key details incomplete. | 2 / 3 |
Workflow Clarity | The front-run → victim → back-run sequence and a profitability threshold are stated, but there is no explicit validate→fix→retry feedback loop for these destructive mainnet operations, which caps workflow clarity at 2. | 2 / 3 |
Progressive Disclosure | The ~105-line skill is a single monolithic file with no bundle files and no one-level-deep references; the tooling/reference lists and code patterns could be split out for clearer navigation. | 2 / 3 |
Total | 8 / 12 Passed |