Content
57%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.
Content is rich in executable, copy-paste-ready code for Hardhat and Foundry testing, which is its main strength. Weaknesses are generic filler sections, duplicated/deprecated content, absence of an explicit testing workflow with validation checkpoints, and a monolithic structure that should be split into reference files.
Suggestions
Delete the generic filler sections ('Instructions' bullets, 'Do not use this skill when' tautologies) and the duplicated gas/mainnet-fork sections; they add tokens without adding knowledge Claude doesn't have.
Add an explicit workflow with checkpoints, e.g.: 1. Write unit tests for core functions, 2. Run 'npx hardhat test' and fix failures, 3. Add fuzz tests for edge cases, 4. Run 'npx hardhat coverage' and only proceed above 90%, 5. Add fork tests for external-protocol interactions.
Split tool-specific detail (hardhat.config.js, Foundry cheatcode reference, CI recipes) into references/ files (e.g., references/foundry.md, references/hardhat-setup.md) and keep SKILL.md as a concise overview with well-signaled links; also replace the deprecated Goerli network config with a current testnet and remove stub snippets that reference undefined variables (addr1, dai in the impersonation example).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly code with little explanatory prose, but it contains pure filler: the generic 'Instructions' section ('Clarify goals, constraints, and required inputs. Apply relevant best practices and validate outcomes'), tautological 'Do not use this skill when' bullets, duplicated gas-optimization and mainnet-forking content across three sections, and deprecated time-sensitive config (Goerli testnet, hardcoded blockNumber 15000000). Mostly efficient but could be tightened, matching anchor 3; not 4 given the amount of removable filler. | 3 / 5 |
Actionability | Mostly executable, copy-paste-ready guidance: complete hardhat.config.js, Mocha/Chai fixtures, Forge test contract with fuzzing and cheatcodes, CI YAML, and CLI commands. Not 5 because several snippets are stubs ('// Test time-dependent functionality', '// Test with real Uniswap contracts', '// Use whale's tokens') and some reference undefined variables (addr1 in the gas test, dai/addr1 in the impersonation snippet, truncated fork URL '...v2/...'). | 4 / 5 |
Workflow Clarity | The section order (setup, unit tests, Foundry, advanced patterns, gas, coverage, verification, CI) implies a rough sequence, but there is no explicit workflow telling Claude how to proceed when testing a contract, and no validation checkpoints (e.g., run tests, confirm coverage before proceeding). Anchor 3 fits: sequence present but checkpoints missing or implicit; not 4 since no checkpoints exist at all. | 3 / 5 |
Progressive Disclosure | Sections are clearly headed and logically grouped, but the entire skill is a ~414-line monolith with no references/, scripts/, or assets/ files; tool-specific content (full hardhat.config.js, large Foundry/Hardhat test samples) belongs in separate reference files with SKILL.md as overview. Between anchors 2 and 3: structure is better than 'minimal' but nothing is offloaded, so 3. | 3 / 5 |
Total | 13 / 20 Passed |