Content
78%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.
A tight, well-structured CLI-wrapper skill: executable safe-start commands, explicit pre-write validation rules, and clean delegation of detail to --help and schema introspection. The main gaps are the absence of a worked example for a common operation and a slightly redundant command table.
Suggestions
Add one worked example of a common read operation (e.g., "gog --readonly --account user@example.com classroom students --course <id> --json --wrap-untrusted") to cover the most likely use case.
Replace the tautological Purpose column entries ("courses | Courses") with a distinguishing detail or flags, or reduce the table to a command list to save tokens.
Add a brief error-recovery note for the safe-start steps, e.g., what to do when "gog auth list --check" reports no authenticated account.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean: it explains nothing Claude already knows, delegates shared auth/safety rules to the sibling skill, and keeps the safe-start block tight. The command table's Purpose column mostly restates the command name ("announcements | Announcements", "courses | Courses"), which is minor padding that could be trimmed — matching anchor 4 rather than the every-token-earns-its-place anchor 5. | 4 / 5 |
Actionability | Three copy-paste-ready commands in the safe-start block and concrete discovery pointers ("gog classroom <command> --help", "gog schema classroom <command> --json") plus the rule "Do not guess command syntax" make the guidance mostly executable. It falls short of anchor 5 because no worked example covers a common case (e.g., listing a course's roster or creating an announcement). | 4 / 5 |
Workflow Clarity | A clear sequence is present (read shared skill → auth check → schema discovery → readonly exploration → operate), and validation for destructive operations is explicit ("Confirm the exact account, object, and mutation before any write or delete", "--dry-run before supported writes", "--no-input in automation"), so the destructive-operation cap does not apply. It misses anchor 5 because there is no error-recovery feedback loop (e.g., what to do when the auth check fails) and the write path is not laid out as an ordered sequence. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines with no bundle files needed; sections are well-organized (Safe start, Commands), the single external reference ("Read ../gog/SKILL.md first for shared auth, output, safety, and live-write rules") is clearly signaled and exactly one level deep, and bulk detail is delegated to --help and schema --json. Per the rubric's simple-skill note, this earns a 5. | 5 / 5 |
Total | 17 / 20 Passed |