Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Code simplification for high-complexity files. Targets deep nesting, copy-paste patterns, god functions, and files over 200 lines.
.claude/skills/hashgraph-online-simplify/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -32% | 0% |
Review all changed files for reuse, quality, and efficiency. Fix any issues found.
Core principle: Identify changes -> Spawn three review agents in parallel -> Fix issues.
Announce at start: "I'm using the simplify skill to review changed code for reuse, quality, and efficiency."
Run git diff (or git diff HEAD if there are staged changes) to see what changed. If there are no git changes, review the most recently modified files that the user mentioned or that were edited earlier in this conversation.
Save the full diff output — it will be passed to each review agent.
Launch all three agents concurrently in a single message. Pass each agent the full diff as context.
| Agent | Role | |-------|------| | code-reuse-reviewer | Finds duplicated logic and missed existing utilities | | code-quality-reviewer | Finds hacky patterns that hurt maintainability | | code-efficiency-reviewer | Finds performance and resource issues |
For Codex, if the Spellbook reviewer agent roles are available, use the namespaced role names:
| Codex Agent Role | Role | |------------------|------| | spellbook-code-reuse-reviewer | Finds duplicated logic and missed existing utilities | | spellbook-code-quality-reviewer | Finds hacky patterns that hurt maintainability | | spellbook-code-efficiency-reviewer | Finds performance and resource issues |
Invocation example (Claude Code):
Use the Agent tool three times in a single message, each with the corresponding subagent_type:
code-reuse-reviewercode-quality-reviewercode-efficiency-reviewerPass each agent a prompt like: > Review the following diff for reuse/quality/efficiency] issues. > > diff > <full diff here> >
Invocation example (Codex):
If the namespaced role names are available in the spawn_agent tool, launch all three agents concurrently with the corresponding agent_type:
spellbook-code-reuse-reviewerspellbook-code-quality-reviewerspellbook-code-efficiency-reviewerPass each agent the same full diff.
For platforms without named agent support: run the three reviews sequentially using each agent's checklist from their definitions.
Wait for all three agents to complete. Aggregate their findings and fix each issue directly. If a finding is a false positive or not worth addressing, note it and move on — do not argue with the finding, just skip it.
Briefly summarize what was fixed (or confirm the code was already clean).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 11,101 | 8,010 | -28% | 1 | 1 | 0% | 817 | 1,103 | +35% | 0 | 0 | — |
case-02 | fail→fail | 11,565 | 3,722 | -68% | 1 | 1 | 0% | 245 | 1,217 | +397% | 0 | 0 | — |
case-03 | fail→fail | 9,250 | 7,627 | -18% | 1 | 1 | 0% | 157 | 929 | +492% | 0 | 0 | — |
case-04 | fail→pass | 12,799 | 6,541 | -49% | 1 | 1 | 0% | 1,722 | 1,537 | -11% | 0 | 0 | — |
case-05 | pass→pass | 6,131 | 3,261 | -47% | 1 | 1 | 0% | 1,165 | 1,255 | +8% | 0 | 0 | — |
case-06 | fail→fail | 13,504 | 10,668 | -21% | 1 | 1 | 0% | 1,779 | 1,663 | -7% | 0 | 0 | — |
case-07 | fail→pass | 19,455 | 8,823 | -55% | 1 | 1 | 0% | 2,325 | 1,238 | -47% | 0 | 0 | — |
case-08 | pass→pass | 15,892 | 8,355 | -47% | 1 | 1 | 0% | 1,454 | 1,234 | -15% | 0 | 0 | — |
case-09 | pass→pass | 13,147 | 9,076 | -31% | 1 | 1 | 0% | 1,638 | 1,342 | -18% | 0 | 0 | — |
case-10 | pass→pass | 17,808 | 4,886 | -73% | 1 | 1 | 0% | 1,706 | 1,321 | -23% | 0 | 0 | — |
case-11 | pass→pass | 11,144 | 11,670 | +5% | 1 | 1 | 0% | 1,467 | 1,736 | +18% | 0 | 0 | — |
case-12 | fail→pass | 13,774 | 8,141 | -41% | 1 | 1 | 0% | 926 | 967 | +4% | 0 | 0 | — |
case-13 | fail→pass | 20,071 | 2,645 | -87% | 1 | 1 | 0% | 2,615 | 1,088 | -58% | 0 | 0 | — |
case-14 | fail→pass | 8,919 | 1,893 | -79% | 1 | 1 | 0% | 1,367 | 936 | -32% | 0 | 0 | — |
case-15 | fail→pass | 6,142 | 2,099 | -66% | 1 | 1 | 0% | 941 | 976 | +4% | 0 | 0 | — |
case-16 | fail→pass | 10,386 | 7,595 | -27% | 1 | 1 | 0% | 1,351 | 1,063 | -21% | 0 | 0 | — |
case-17 | fail→pass | 15,747 | 8,366 | -47% | 1 | 1 | 0% | 2,344 | 1,239 | -47% | 0 | 0 | — |
case-18 | pass→pass | 9,329 | 8,026 | -14% | 1 | 1 | 0% | 1,371 | 1,058 | -23% | 0 | 0 | — |
case-19 | fail→pass | 12,701 | 7,990 | -37% | 1 | 1 | 0% | 1,349 | 1,093 | -19% | 0 | 0 | — |
case-20 | fail→pass | 16,883 | 7,273 | -57% | 1 | 1 | 0% | 1,977 | 1,046 | -47% | 0 | 0 | — |
case-21 | pass→pass | 18,488 | 8,486 | -54% | 1 | 1 | 0% | 2,118 | 2,065 | -3% | 0 | 0 | — |
case-22 | fail→fail | 11,120 | 32,224 | +190% | 1 | 1 | 0% | 1,558 | 6,147 | +295% | 0 | 0 | — |
case-23 | fail→fail | 23,748 | 9,172 | -61% | 1 | 1 | 0% | 3,700 | 1,131 | -69% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 21 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +43 percentage points is the difference between those two pass rates over the 21 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.