Content
80%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is an exemplar of token-efficient expert guidance: terse, non-obvious, and free of padding, with concrete checklists for EVM, oracle, and upgrade analysis. Its main weakness is that the verification flow (scanner finding → manual confirmation → harness reproduction) is implied rather than sequenced with explicit checkpoints. Guidance on deferring implementation detail to other technique skills is explicit and deliberate.
Suggestions
Make the implicit verification flow explicit, e.g., a short sequence: 'pattern candidate → confirm caller/modifier/storage state → Foundry harness against pinned fork → report with preserved flags and inputs.'
Add a brief validation checkpoint per analysis area (e.g., 'do not cite impact without a reproduced harness or an explicit assumption list') so evidentiary limits become gates rather than cautions.
Name the specific technique/contracts skills being deferred to (or their paths) so the handoff is unambiguous.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, expert-level, and free of padding — it never explains basics (what Slither is, what a proxy does) and every sentence adds non-obvious guidance, e.g., 'The legacy slither_ingest backend is separate from the engagement KG; neither scanner ingestion nor a graph edge proves exploitability.' This matches 'Lean and efficient; assumes Claude's competence; every token earns its place.' | 5 / 5 |
Actionability | For an instruction-only skill the guidance is concrete: exact oracle checklist items ('price source, decimal handling, update cadence, liquidity, TWAP window and consumers') and precise harness requirements ('preserve compiler flags, fork block and test inputs'). It is not a 5 because there are no copy-paste commands or example harnesses, and not a 3 because the checklists are specific and executable in substance rather than pseudocode-like hints. | 4 / 5 |
Workflow Clarity | An implicit verify-then-claim sequence exists (pattern candidate → confirm caller/modifier/state → Foundry harness → evidence), but it is never laid out as explicit steps with checkpoints, matching 'sequence present but checkpoints missing or implicit.' It is not a 4 because no ordered procedure or validation gate is stated; the skill itself notes it deliberately does not define an engagement stage order. | 3 / 5 |
Progressive Disclosure | The body is under 50 lines with well-organized sections and no need for external references — no references/, scripts/, or assets/ directories exist, and pointers to other technique skills are clearly signaled in prose. This matches the rubric's simple-skill case: 'under 50 lines with no need for external references... score 5 with just well-organized sections.' | 5 / 5 |
Total | 17 / 20 Passed |