Content
67%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 a clear, actionable shell-first workflow with concrete commands, sensible guardrails, and a properly deferred reference file. Its main weakness is redundancy — the same warnings and reference pointers repeated across sections — plus the absence of an explicit validate/fix/retry loop in the build-debug workflow.
Suggestions
State the raw-executable warning for GUI apps once (in Guardrails) and remove the duplicates in the Quick Start bullets and workflow step 3 to reduce token cost.
Cite references/run-button-bootstrap.md once in the References section (or once in the workflow step where the script shape is defined) instead of three times.
Add an explicit failure-recovery checkpoint after step 5, e.g. 'If the build fails: fix the top blocker, re-run ./script/build_and_run.sh, and only proceed when it launches' to close the validate-fix-retry loop.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is directive and command-dense with no explanation of concepts Claude already knows, but it is noticeably redundant: the warning against raw executable launch for GUI apps appears three times (Quick Start, workflow step 3, Guardrails), the kill/build/run framing repeats across sections, and the pointer to references/run-button-bootstrap.md is given three times. It fits 'mostly efficient but could be tightened' rather than 4, since the redundancy exceeds minor over-explanation. | 3 / 5 |
Actionability | Concrete executable commands appear throughout: git rev-parse --is-inside-work-tree, find . -name '*.xcworkspace' ..., xcodebuild -list -workspace <workspace>, pgrep -x <AppName>, /usr/bin/open -n <bundle>, exact Info.plist keys, and exact flag names for the run script. It stops short of 5 because the actual script body is deferred to the reference file rather than being copy-paste ready in the skill. | 4 / 5 |
Workflow Clarity | An 8-step clearly sequenced workflow with most checkpoints present: --verify with pgrep for process confirmation, step 6 failure classification (compiler, linker, signing, etc.), the git init nested-repo guardrail, and the conditional to update an existing environment.toml rather than duplicate it. Not 5 because there is no explicit fix-and-re-validate loop after a failed build, so error-recovery feedback is implicit. | 4 / 5 |
Progressive Disclosure | A single one-level-deep reference (references/run-button-bootstrap.md, verified to exist) is well signaled both inline ('canonical source... Do not fork a second authoritative snippet') and in a dedicated References section, with the body organized under clear headers. It misses 5 because the body duplicates reference-level detail (Info.plist keys, exact flag behavior) instead of fully deferring it, and repeats the pointer three times. | 4 / 5 |
Total | 15 / 20 Passed |