Content
86%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 single-purpose skill with executable commands and explicit validation checkpoints for a non-destructive, read-only-by-default operation. The main gaps are minor: JSON output parsing is described rather than demonstrated, and no retry feedback loop is specified.
Suggestions
Add a one-line concrete example of parsing the JSON payload (e.g. a jq snippet checking a success field) so the 'decode failure from a zero exit' guidance is fully executable rather than only described.
Specify a small feedback loop for the run step, e.g. on a reported payload error or timeout, surface the error to the user and optionally retry once with a corrected prompt rather than only stopping.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence: direct commands ("command -v openclaw", the openclaw invocation) with no padding or explanations of concepts Claude already knows, matching anchor 5; the non-obvious note about decoding failure from a zero exit earns its place rather than being filler. | 5 / 5 |
Actionability | It gives a fully executable, copy-paste-ready command ("openclaw agent --local --agent main --message \"<prompt>\" --json --timeout 600") plus an availability check, fitting anchor 4; it stops short of anchor 5 because parsing the JSON payload for encoded failures is described ("Parse the JSON payload") rather than shown with concrete code. | 4 / 5 |
Workflow Clarity | There is a clear numbered sequence (set cwd/timeout, check availability, run) with explicit validation checkpoints (availability check, decode failure-from-success, stop on missing auth), fitting anchor 4; it lacks the explicit fix-and-retry feedback loop that would warrant anchor 5. | 4 / 5 |
Progressive Disclosure | The body is under 50 lines, needs no external references, and is organized into clearly signaled sections (Run, Authentication And Permissions), so per the simple-skill guideline it matches anchor 5; no bundle directories exist to verify further. | 5 / 5 |
Total | 18 / 20 Passed |