Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Expert in Confluence operations using Atlassian MCP. Use when the user says "search Confluence", "create a Confluence page", "update a page", "find documentation in Confluence", "list spaces", or "add a comment to a page". Do NOT use for Jira issues, general web search, or local file creation.
.claude/skills/tech-leads-club-confluence-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 79% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 70% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 55% | 0% |
You are an expert in using Atlassian MCP tools to interact with Confluence.
Use this skill when the user asks to:
Project Detection Strategy (Automatic):
When you activate this skill:
https://example.atlassian.net/)"Cloud ID format:
https://example.atlassian.net/)getAccessibleAtlassianResourcesUse search (Rovo Search) first - it's the most efficient way:
search("natural language query about the content")Depending on what you have:
fetch(ari)getConfluencePage(cloudId, pageId)getConfluenceSpaces(cloudId, keys=["SPACE_KEY"])getPagesInConfluenceSpace(cloudId, spaceId)createConfluencePage(
cloudId,
spaceId="123456",
title="Page Title",
body="# Markdown Content\n\n## Section\nContent here..."
)Always use Markdown in the body field — never HTML.
updateConfluencePage(
cloudId,
pageId="123456",
title="Updated Title",
body="# Updated Markdown Content\n\n..."
)Always use Markdown in the body field — never HTML.
body fieldsearch first before other lookup methodsari:cloud:confluence:site-id:page/page-idUser: "Find the API documentation page and add a new section"
1. search("API documentation")
2. getConfluencePage(cloudId, pageId="found-id")
3. updateConfluencePage(
cloudId,
pageId="found-id",
title="API Documentation",
body="# API Documentation\n\n## Existing Content\n...\n\n## New Section\nNew content here..."
)User: "Create a new architecture decision record"
1. getConfluenceSpaces(cloudId, keys=["TECH"])
2. createConfluencePage(
cloudId,
spaceId="space-id-from-step-1",
title="ADR-001: Use Microservices Architecture",
body="# ADR-001: Use Microservices Architecture\n\n## Status\nAccepted\n\n## Context\n...\n\n## Decision\n...\n\n## Consequences\n..."
)User: "What's in our onboarding documentation?"
1. search("onboarding documentation")
2. getConfluencePage(cloudId, pageId="id-from-results")
3. Summarize the content for the userWhen creating or updating pages, use well-structured Markdown:
markdown# Main Title ## Introduction Brief overview of the topic. ## Sections Organize content logically with: - Clear headings (##, ###) - Bullet points for lists - Code blocks for examples - Tables when appropriate ## Key Points - Point 1 - Point 2 - Point 3 ## Next Steps 1. Step 1 2. Step 2 3. Step 3
body| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→pass | 5,955 | 13,070 | +119% | 1 | 1 | 0% | 1,077 | 1,929 | +79% | 0 | 0 | — |
case-01 | fail→fail | 24,204 | 5,605 | -77% | 1 | 1 | 0% | 1,087 | 1,529 | +41% | 0 | 0 | — |
case-02 | fail→fail | 13,216 | 3,646 | -72% | 1 | 1 | 0% | 2,203 | 1,913 | -13% | 0 | 0 | — |
case-03 | fail→fail | 6,439 | 5,476 | -15% | 1 | 1 | 0% | 1,065 | 1,655 | +55% | 0 | 0 | — |
case-04 | pass→pass | 9,701 | 7,230 | -25% | 1 | 1 | 0% | 1,762 | 2,492 | +41% | 0 | 0 | — |
case-05 | fail→pass | 11,091 | 7,401 | -33% | 1 | 1 | 0% | 2,117 | 2,387 | +13% | 0 | 0 | — |
case-07 | fail→pass | 53,478 | 4,047 | -92% | 1 | 1 | 0% | 1,215 | 2,060 | +70% | 0 | 0 | — |
case-08 | fail→fail | 4,652 | 10,197 | +119% | 1 | 1 | 0% | 654 | 1,627 | +149% | 0 | 0 | — |
case-09 | fail→fail | 54,700 | 6,941 | -87% | 1 | 1 | 0% | 1,043 | 1,555 | +49% | 0 | 0 | — |
case-10 | fail→pass | 7,886 | 1,901 | -76% | 1 | 1 | 0% | 1,503 | 1,669 | +11% | 0 | 0 | — |
case-11 | fail→pass | 6,012 | 19,832 | +230% | 1 | 1 | 0% | 1,132 | 1,750 | +55% | 0 | 0 | — |
case-12 | pass→pass | 8,470 | 4,971 | -41% | 1 | 1 | 0% | 1,457 | 2,211 | +52% | 0 | 0 | — |
case-13 | fail→pass | 8,628 | 2,873 | -67% | 1 | 1 | 0% | 1,415 | 1,796 | +27% | 0 | 0 | — |
case-14 | pass→fail | 8,680 | 19,581 | +126% | 1 | 1 | 0% | 1,617 | 2,491 | +54% | 0 | 0 | — |
case-15 | pass→pass | 8,378 | 3,825 | -54% | 1 | 1 | 0% | 832 | 1,956 | +135% | 0 | 0 | — |
case-16 | fail→fail | 16,542 | 2,107 | -87% | 1 | 1 | 0% | 2,340 | 1,673 | -29% | 0 | 0 | — |
case-17 | pass→pass | 5,696 | 2,134 | -63% | 1 | 1 | 0% | 1,011 | 1,689 | +67% | 0 | 0 | — |
case-18 | pass→pass | 9,867 | 4,029 | -59% | 1 | 1 | 0% | 1,733 | 1,944 | +12% | 0 | 0 | — |
case-19 | pass→pass | 9,204 | 8,697 | -6% | 1 | 1 | 0% | 1,581 | 2,696 | +71% | 0 | 0 | — |
case-20 | fail→pass | 8,849 | 3,138 | -65% | 1 | 1 | 0% | 1,650 | 1,915 | +16% | 0 | 0 | — |
case-21 | fail→pass | 3,397 | 9,572 | +182% | 1 | 1 | 0% | 546 | 2,104 | +285% | 0 | 0 | — |
case-22 | fail→pass | 3,699 | 5,077 | +37% | 1 | 1 | 0% | 535 | 2,194 | +310% | 0 | 0 | — |
case-23 | fail→pass | 6,106 | 5,391 | -12% | 1 | 1 | 0% | 864 | 2,092 | +142% | 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 18 counted toward the lift figure. The other 5 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 +39 percentage points is the difference between those two pass rates over the 18 comparable cases. 1 case got worse with the skill loaded, and it is 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.