Content
82%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 high-quality, information-dense skill body: nearly every line states a project-specific constraint or an executable command, and validation is enforced with explicit completion criteria. Minor room remains in tightening a few run-on passages and splitting the detailed CDP inspection recipes into a reference file.
Suggestions
Split the installed-app and dev-window CDP inspection recipes (ports, profile lock, user-data-dir workaround, pid-stop trick) into a references file, keeping in SKILL.md only the one-line trigger for when to load it.
Tighten the multi-clause explanatory sentences (e.g., the 'A native-looking change is the usual place this gets skipped' sentence and the dev-window paragraph) into shorter directive statements, keeping the rationale as a brief trailing clause.
Name the actual command(s) for 'Run focused Electron tests and package checks' in the Validation section so the final checkpoint is as executable as the rest of the guidance.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes Claude's competence — it explains no general Electron concepts and nearly every line carries project-specific facts (e.g., 'windowsHide reliably controls only the directly spawned process'). A few long run-on passages ('A native-looking change is the usual place this gets skipped: a shutdown order, an updater handoff...') and the multi-clause dev-window paragraph could be tightened, matching anchor 4 rather than the fully lean anchor 5. | 4 / 5 |
Actionability | Guidance is copy-paste ready and covers the common cases: exact spawn flags ('windowsHide: true', 'stdio: \'ignore\'', 'powershell.exe -NoProfile -NonInteractive -WindowStyle Hidden -EncodedCommand ...'), a launchable command ('.../MacOS/OpenChamber --remote-debugging-port=9222'), exact env vars, ports, file paths, and the pid-based stop trick. This matches anchor 5's fully executable, common-case coverage. | 5 / 5 |
Workflow Clarity | The flow is context-first (Required Context with a completion criterion) and closes with an explicit Validation section containing hard checkpoints ('completion requires both HMR development and bundled UI validation', 'command success alone is insufficient'). It stops short of anchor 5 because steps are organized as topical rules rather than a sequenced workflow, and there are no explicit validate-fix-retry feedback loops. | 4 / 5 |
Progressive Disclosure | No bundle files exist; sections are clearly headed and external material is referenced one level deep with clear signals (packages/electron/README.md, the ui-api-decoupling skill, openchamber-change-discipline's references/updater-testing.md). The extensive inline CDP debugging recipes in the Validation section are content that could plausibly live in a separate reference file, which keeps this at anchor 4 rather than 5. | 4 / 5 |
Total | 17 / 20 Passed |