Content
78%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 well-structured, actionable skill body with strong progressive disclosure: platform detail is correctly split into five real reference files, and the workflows include genuine validation checkpoints and error-recovery fallbacks. The main weaknesses are a subtly broken printf example in section A, some repeated platform-pointer prose, and no post-install verification step.
Suggestions
Fix the section A snippet: `printf '%s'` does not interpret `\n` in its argument, so the piped body contains literal backslash-n. Use a quoted heredoc instead, e.g. `memsearch skills add --name "<slug>" --description "..." --body-file - <<'EOF'`.
State the platform-reference location once (the intro list) and drop the three later "see your platform reference file" repetitions in B, D, and Install paths to save tokens.
Add a post-install verification step in section B (e.g. re-run `memsearch skills status` or check the installed path) so the install flow closes with an explicit confirmation rather than only a restart reminder.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient and domain-specific (MemSearch's memory layers and skills status semantics are genuinely unknown to Claude), but the platform-reference pointer is repeated four times (intro list, "Replace <platform> with your platform key prefix (see the reference file)", "See your platform reference file for the exact...commands", "See your platform reference file for the platform-specific skill directory") and concepts like the git-tracked store are restated, so it sits at anchor 4 rather than 5. | 4 / 5 |
Actionability | Concrete, specific commands throughout — "memsearch skills status", "memsearch skills list", "memsearch skills install <name> --path <...>", real config keys like "plugins.<platform>.memory_to_skill.min_occurrences". However the flagship snippet "printf '%s' \"## <title>\\n\\n1. ...\" | memsearch skills add ... --body-file -" is not copy-paste ready: printf '%s' does not interpret \n in its argument, so the piped body would contain literal backslash-n instead of newlines. That keeps it below anchor 5. | 4 / 5 |
Workflow Clarity | Each route (A/B/C/D) has a clear sequence with real checkpoints: "installing copies the candidate as-is, so this is the last chance to catch a wrong step", an interactive confirm-destination gate, "Only if that command fails (unknown format) fall back to reading the raw file directly", and "never fabricate". Minor gaps keep it at 4: section A's flow is prose rather than numbered steps, and there is no post-install verification that the skill landed (only a reminder to restart). | 4 / 5 |
Progressive Disclosure | The body is a clear overview that keeps platform-specific detail out of SKILL.md behind well-signaled, one-level-deep references; all five files named in the body (references/claude-code.md, codex.md, openclaw.md, opencode.md, dsh.md) exist in the bundle, matching the platform list exactly. | 5 / 5 |
Total | 17 / 20 Passed |