Content
93%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.
This is a well-engineered skill body: a lean, copy-paste-ready CLI workflow with useful non-obvious rules (stderr redirection, telemetry user-agent), dynamic fallbacks (actors search when the index has no match), and an exemplary progressive-disclosure structure — every referenced file exists, is substantive, and is one level deep. The only gap is a missing explicit failure-recovery loop (what to do on FAILED runs), which keeps workflow clarity at 4 rather than 5.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and operational: three terse global rules ('--json', user-agent telemetry, '2>/dev/null' because 'stderr contains progress messages that break JSON parsers'), copy-paste commands, and output-field mappings ('.id', '.defaultDatasetId', '.stats.durationMillis') — it never explains concepts Claude already knows (no 'what is an Actor', no library background). Every token earns its place, matching the anchor-5 example's bare heading-plus-command pattern; not 4, since I found no padded or over-explanatory passage to trim. | 5 / 5 |
Actionability | Fully executable copy-paste commands throughout: 'apify actors call "ACTOR_ID" --input-file input.json --user-agent ... --json 2>/dev/null', with guidance on inline vs file inputs, result fetching ('apify datasets get-items DATASET_ID --format csv'), non-blocking runs with polling ('apify runs info RUN_ID ... check .status for SUCCEEDED'), and delivery conventions (filename format 'YYYY-MM-DD_descriptive-name.csv', console links). Only placeholders like ACTOR_ID are left to the reader, which is appropriate; matches anchor 5's 'copy-paste ready... covers the common cases'. | 5 / 5 |
Workflow Clarity | A clear four-step sequence (select Actor -> fetch schema + check gotchas -> configure/run -> deliver), with checkpoints: gotchas review before PPE runs, 'Skip user preferences for simple lookups', and status polling ('check .status for SUCCEEDED') before fetching results. It falls short of anchor 5 because there is no explicit error-recovery feedback loop — e.g. what to do when a run's status is FAILED or TIMED-OUT beyond the Troubleshooting pointer to gotchas.md; clearly above anchor 3, whose 'test the output' step is implicit, since status verification and pre-run cost checks are explicit. | 4 / 5 |
Progressive Disclosure | The body is a clean overview that delegates all detail through well-signaled, one-level-deep references, all verified to exist and be substantive: 'references/actor-index.md' for Actor selection, 'references/gotchas.md' for pitfalls, and a task-keyword routing table of 14 'references/workflows/*.md' guides (lead-generation.md, competitive-intel.md, etc.). This is the anchor-5 pattern exactly (concise overview + clearly signaled reference files with easy navigation); nothing is inlined that belongs in a separate file, and no reference nests deeper than one level. | 5 / 5 |
Total | 19 / 20 Passed |