Content
81%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 exemplary operational skill: concrete executable commands, an explicit structured-first decision hierarchy, and thorough verification loops with error recovery. Its only real weakness is repetition — several rules and path setup lines are restated across sections, which costs tokens without adding clarity.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body teaches only operational specifics Claude would not know, but rules repeat noticeably — screenshot-freshness appears in four places (operating model, standing rules, control flow, terminal loop), boundary rules recur across three sections, and the DESKTOP= path assignment is restated in every code block — leaving more than minor tightening to do. | 3 / 5 |
Actionability | Every section gives copy-paste-ready bash: exact absolute script paths, named subcommands (state --json, wait-window, calc-set-cell FILE SHEET CELL VALUE), a working shell fallback one-liner for CLI launches, and a complete heredoc sequence example — fully executable coverage of the common cases. | 5 / 5 |
Workflow Clarity | The 8-step decision hierarchy, 7 standing rules, and 6-step terminal verification loop are explicitly sequenced with validation checkpoints, error-recovery paths (report readiness instead of retrying clicks; fallback CLI on 'command not found'), and stale-evidence rejection rules — a genuine validate/fix/retry feedback loop. | 5 / 5 |
Progressive Disclosure | A well-sectioned single file with clear headers whose bundled scripts/desktopctl.sh is consistently referenced one level deep; minor gaps are the absence of relative pointers into the bundle (calc_set_cell.py is never surfaced) and ~35-line sections (terminal verification, app map) that could live in reference files. | 4 / 5 |
Total | 17 / 20 Passed |