Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Perform Jira and Confluence operations using the Atlassian CLI (acli). Use for issues, sprints, boards, pages, spaces, and project management.
.claude/skills/abdullahkhawer-jira-and-confluence-cli/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 144% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -23% | 0% |
Use the instance and rules below when constructing all acli commands.
Set your Atlassian site URL before using this skill:
| Setting | Value | |---|---| | Atlassian Site | <your-site>.atlassian.net |
Replace <your-site>.atlassian.net with your actual Atlassian site domain throughout this skill.
If not already set up, inform the user that setup is required and ask them to complete the following steps manually. Do not perform these steps yourself.
bashbrew tap atlassian/homebrew-acli brew install acli
bashacli auth login
This opens a browser window at https://api.atlassian.com/oauth2/authorize/.... Click Accept to grant the Atlassian CLI the required OAuth scopes. Once authorized, return to the terminal and select your site from the list when prompted.
--limit 100 (or use --paginate if supported) to ensure complete results are returned.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 6,395 | 2,747 | -57% | 1 | 1 | 0% | 1,254 | 803 | -36% | 0 | 0 | — |
case-06 | fail→pass | 5,311 | 3,827 | -28% | 1 | 1 | 0% | 866 | 1,091 | +26% | 0 | 0 | — |
case-02 | fail→pass | 5,848 | 5,320 | -9% | 1 | 1 | 0% | 975 | 1,438 | +47% | 0 | 0 | — |
case-03 | fail→pass | 5,187 | 5,823 | +12% | 1 | 1 | 0% | 598 | 1,461 | +144% | 0 | 0 | — |
case-04 | fail→pass | 7,627 | 3,723 | -51% | 1 | 1 | 0% | 1,391 | 1,068 | -23% | 0 | 0 | — |
case-05 | fail→pass | 3,985 | 6,354 | +59% | 1 | 1 | 0% | 629 | 1,491 | +137% | 0 | 0 | — |
case-07 | fail→pass | 3,440 | 3,030 | -12% | 1 | 1 | 0% | 629 | 954 | +52% | 0 | 0 | — |
case-08 | fail→pass | 6,720 | 3,820 | -43% | 1 | 1 | 0% | 882 | 1,109 | +26% | 0 | 0 | — |
case-09 | fail→fail | 5,322 | 7,991 | +50% | 1 | 1 | 0% | 663 | 1,858 | +180% | 0 | 0 | — |
case-10 | fail→pass | 7,118 | 3,554 | -50% | 1 | 1 | 0% | 1,401 | 1,081 | -23% | 0 | 0 | — |
case-11 | fail→pass | 7,871 | 4,252 | -46% | 1 | 1 | 0% | 1,546 | 1,145 | -26% | 0 | 0 | — |
case-12 | pass→pass | 6,386 | 10,633 | +67% | 1 | 1 | 0% | 1,203 | 920 | -24% | 0 | 0 | — |
case-13 | fail→pass | 5,909 | 3,030 | -49% | 1 | 1 | 0% | 1,048 | 858 | -18% | 0 | 0 | — |
case-14 | fail→pass | 25,372 | 3,597 | -86% | 1 | 1 | 0% | 1,085 | 985 | -9% | 0 | 0 | — |
case-23 | pass→pass | 12,020 | 2,455 | -80% | 1 | 1 | 0% | 652 | 870 | +33% | 0 | 0 | — |
case-15 | fail→pass | 7,501 | 1,715 | -77% | 1 | 1 | 0% | 1,480 | 654 | -56% | 0 | 0 | — |
case-16 | fail→pass | 7,462 | 3,986 | -47% | 1 | 1 | 0% | 1,366 | 1,104 | -19% | 0 | 0 | — |
case-17 | fail→pass | 9,722 | 3,660 | -62% | 1 | 1 | 0% | 1,928 | 1,113 | -42% | 0 | 0 | — |
case-18 | fail→pass | 6,183 | 5,277 | -15% | 1 | 1 | 0% | 1,163 | 1,375 | +18% | 0 | 0 | — |
case-19 | fail→pass | 8,015 | 3,542 | -56% | 1 | 1 | 0% | 1,308 | 1,015 | -22% | 0 | 0 | — |
case-20 | fail→pass | 5,156 | 3,590 | -30% | 1 | 1 | 0% | 978 | 1,056 | +8% | 0 | 0 | — |
case-21 | pass→pass | 3,696 | 4,318 | +17% | 1 | 1 | 0% | 737 | 1,165 | +58% | 0 | 0 | — |
case-22 | pass→pass | 8,831 | 7,531 | -15% | 1 | 1 | 0% | 1,904 | 1,796 | -6% | 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. The headline lift of +78 percentage points is the difference between those two pass rates over the 23 comparable cases.
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.