Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Simple operations on user-provided text files including summarization.
.claude/skills/microck-user-file-ops/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 173% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -59% | 0% |
| case-23 | ✓→✗ | ▼ Worse | -59% | 0% |
| case-20 | ✓→✓ | ＝ Same ✓ | 245% | 0% |
| case-21 | ✓→✓ | ＝ Same ✓ | -36% | 0% |
Overview
Summarize text files that come from the user or from other skills. This skill can compute basic statistics (lines, words, bytes) and capture a short preview of the file.
User-provided files are typically exposed under work/inputs/ (for example, when a host directory is mounted as inputs). Files produced by other skills are usually written under out/ and can be summarized directly from there.
Examples
1) Summarize a text file already present in the workspace
Command:
bash scripts/summarize_file.sh \ work/inputs/example.txt \ out/example_summary.txt
2) Summarize a different file
Command:
bash scripts/summarize_file.sh \ work/inputs/notes.txt \ out/notes_summary.txt
3) Summarize a file produced by another skill
Command:
bash scripts/summarize_file.sh \ out/sample_fib.txt \ out/sample_fib_summary.txt
Output Files
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 2,927 | 5,045 | +72% | 1 | 1 | 0% | 307 | 538 | +75% | 0 | 0 | — |
case-07 | fail→fail | 7,576 | 4,785 | -37% | 1 | 1 | 0% | 563 | 424 | -25% | 0 | 0 | — |
case-08 | fail→fail | 9,752 | 4,758 | -51% | 1 | 1 | 0% | 1,994 | 450 | -77% | 0 | 0 | — |
case-02 | fail→fail | 2,971 | 4,866 | +64% | 1 | 1 | 0% | 414 | 503 | +21% | 0 | 0 | — |
case-03 | fail→fail | 13,563 | 4,810 | -65% | 1 | 1 | 0% | 2,466 | 503 | -80% | 0 | 0 | — |
case-04 | fail→fail | 14,742 | 5,351 | -64% | 1 | 1 | 0% | 3,080 | 554 | -82% | 0 | 0 | — |
case-05 | fail→pass | 5,169 | 13,169 | +155% | 1 | 1 | 0% | 882 | 2,409 | +173% | 0 | 0 | — |
case-06 | fail→fail | 6,457 | 4,920 | -24% | 1 | 1 | 0% | 1,359 | 518 | -62% | 0 | 0 | — |
case-09 | fail→fail | 6,366 | 5,140 | -19% | 1 | 1 | 0% | 302 | 488 | +62% | 0 | 0 | — |
case-10 | fail→fail | 17,721 | 4,814 | -73% | 1 | 1 | 0% | 2,558 | 457 | -82% | 0 | 0 | — |
case-11 | fail→fail | 6,334 | 4,607 | -27% | 1 | 1 | 0% | 248 | 471 | +90% | 0 | 0 | — |
case-12 | fail→fail | 4,040 | 5,015 | +24% | 1 | 1 | 0% | 164 | 565 | +245% | 0 | 0 | — |
case-13 | fail→fail | 13,119 | 4,356 | -67% | 1 | 1 | 0% | 2,714 | 458 | -83% | 0 | 0 | — |
case-14 | fail→fail | 10,183 | 4,880 | -52% | 1 | 1 | 0% | 1,444 | 520 | -64% | 0 | 0 | — |
case-15 | fail→fail | 2,953 | 4,891 | +66% | 1 | 1 | 0% | 320 | 481 | +50% | 0 | 0 | — |
case-16 | fail→fail | 20,745 | 4,335 | -79% | 1 | 1 | 0% | 3,960 | 502 | -87% | 0 | 0 | — |
case-17 | fail→fail | 6,438 | 5,025 | -22% | 1 | 1 | 0% | 1,253 | 474 | -62% | 0 | 0 | — |
case-18 | fail→fail | 8,691 | 4,547 | -48% | 1 | 1 | 0% | 548 | 416 | -24% | 0 | 0 | — |
case-19 | fail→fail | 18,359 | 2,676 | -85% | 1 | 1 | 0% | 3,985 | 523 | -87% | 0 | 0 | — |
case-20 | pass→pass | 7,282 | 11,847 | +63% | 1 | 1 | 0% | 651 | 2,245 | +245% | 0 | 0 | — |
case-21 | pass→pass | 5,852 | 2,761 | -53% | 1 | 1 | 0% | 936 | 598 | -36% | 0 | 0 | — |
case-22 | pass→fail | 22,770 | 4,507 | -80% | 1 | 1 | 0% | 1,013 | 415 | -59% | 0 | 0 | — |
case-23 | pass→fail | 10,560 | 6,480 | -39% | 1 | 1 | 0% | 1,448 | 593 | -59% | 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 5 counted toward the lift figure. The other 18 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. A headline lift is not published for this run.
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.