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.
The body delivers genuinely useful, unknowable reference data (Actor IDs, use-case mappings) and concrete executable commands for schema fetching, organized in a clear 5-step workflow. Its main weaknesses are the missing run_actor.js script the whole execution depends on, no output-validation step for batch scraping runs, and a large inline Actor catalog that should live in a separate reference file.
Suggestions
Ship the run_actor.js script in the bundle (e.g., scripts/run_actor.js) and include one complete worked example showing how to build JSON_INPUT from a fetched schema (e.g., a compass/crawler-google-places run with a real search string and maxCrawledPlaces).
Add a validation checkpoint between running the Actor and summarizing results: check the output file is non-empty, row count matches expectations, and retry with reduced input or increased --timeout on failure — this also lifts the workflow-clarity batch cap.
Move the 55-Actor catalog tables to a references/actors.md file, keeping only the use-case selection table and a pointer in SKILL.md, and consolidate the three near-identical run commands into one with documented optional --output/--format flags.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and high-value: Actor ID tables, use-case mappings, and concrete commands that Claude could not know otherwise. Minor trimmable padding remains — the frontmatter description is repeated at the top, "(No need to check it upfront)", and three near-identical bash blocks for quick/CSV/JSON output that could be one command with optional flags. Not score 5 because of this redundancy. | 4 / 5 |
Actionability | The mcpc search and schema-fetch commands are fully executable, but the core execution step depends on `${CLAUDE_PLUGIN_ROOT}/reference/scripts/run_actor.js`, a script that does not exist anywhere in the skill bundle (no scripts/ directory), and no concrete example shows how to construct `JSON_INPUT`. This is more than a minor gap, matching anchor 3 rather than 4. | 3 / 5 |
Workflow Clarity | The 5-step sequence is clearly laid out with a copyable progress checklist and schema-fetching before execution, but scraping runs are batch operations with no output validation/verification step before reporting results — error handling is reactive only ("Run FAILED - Ask user to check Apify console link"). Per the batch-operation cap, workflow clarity cannot exceed 3 without validation. Not score 2 because the sequence itself is well defined. | 3 / 5 |
Progressive Disclosure | The single SKILL.md is well-sectioned with headers and tables, but ~100 lines of Actor catalog that belongs in a reference file are inlined, and the only file reference (run_actor.js) points to a path that does not exist in the bundle. Some structure with content that should be separate inline and a broken reference — anchor 3, not 4 given the missing script target. | 3 / 5 |
Total | 13 / 20 Passed |