Content
61%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 well-structured with genuinely executable quick-start commands for the website pipeline and clearly signaled pointers to per-pipeline documentation. It is weakened by ~25% promotional/contributory padding that wastes token budget, the absence of validation checkpoints in the build-deploy workflow, and concrete steps for only one of the seven pipelines.
Suggestions
Remove or trim the 'Contributing Back', 'For OpenClaw Agents', and token/link promotional sections — they consume token budget without aiding task execution.
Add explicit validation checkpoints to the workflow, e.g. 'After npm run build, confirm zero errors before deploying' and 'Verify the returned previewUrl responds before reporting success', plus error-recovery guidance for build/deploy failures.
Provide at least minimal concrete first commands for each pipeline (as done for websites) instead of only 'Follow <pipeline>/CLAUDE.md' pointers.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The core build guidance (pipeline table, quick-start commands, deploy steps) is efficient, but roughly a quarter of the body is non-skill padding: the ~40-line 'Contributing Back' section, 'For OpenClaw Agents', and promotional links ('Token Launchpad', '$FACTORY Token' Jupiter swap URL). This is more than the minor trimming of a 4 but not the pervasive over-explanation of a 2. | 3 / 5 |
Actionability | The website path is fully executable and copy-paste ready ('git clone ...', 'npx create-next-app@latest . --typescript --tailwind --app --src-dir ...', 'npx vercel --prod --yes', 'bash deploy.sh'), and the design system gives concrete values. It is not a 5 because six of the seven pipelines get only a 'Follow app-factory/CLAUDE.md' pointer with no concrete steps in the skill itself. | 4 / 5 |
Workflow Clarity | Quick Start presents a clear 4-step sequence (clone → build → deploy → claim), but validation checkpoints are missing or implicit: no step verifies the build succeeded or the deployment is live, and no error-recovery guidance exists. 'LOCAL_RUN_PROOF_GATE' is described as an architectural feature rather than a workflow step, so the content fits the 'checkpoints missing' anchor rather than the 'most checkpoints present' one. | 3 / 5 |
Progressive Disclosure | No bundle files exist, so structure is scored on the body alone: clear section headers, a pipeline summary table, and one-level-deep, clearly signaled references to per-pipeline CLAUDE.md files ('Follow website-pipeline ... Follow app-factory/CLAUDE.md for the full 10-stage pipeline'). Minor gaps — the contributing and design-system content is inline and references point into an external repo rather than a local bundle — keep it below 5. | 4 / 5 |
Total | 14 / 20 Passed |