Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Configure MCP servers for n8n development in OpenCode. Use when the user says /setup-mcps or asks to set up MCP servers for n8n.
.claude/skills/n8n-io-n8n-setup-mcps/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -72% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -16% | 0% |
Configure commonly used MCP servers for n8n engineers.
bashclaude mcp list
Parse the output and match by URL/command, not server name (users may have used different names). The URLs to check for:
mcp.linear.appmcp.notion.comctx7 or context7-mcpmcp.figma.comSkip any MCP whose URL/command is already present (regardless of scope or name).
AskUserQuestion with multiSelect: true.Only show MCPs that are not already configured. If all are already configured, inform the user and skip the menu.
| Option | Label | Description | |--------|-------|-------------| | Linear | Linear | Linear ticket management (HTTP, OAuth — opens browser to authenticate) | | Notion | Notion | Notion workspace integration (HTTP, OAuth — opens browser to authenticate) | | Context7 | Context7 | Library documentation lookup (OAuth setup via CLI) | | Figma | Figma | Figma design integration (HTTP, OAuth — opens browser to authenticate) |
a. Ask the user via AskUserQuestion: "Where should {MCP name} be installed?"
b. Run the install command for that MCP with the chosen scope c. Then move to the next MCP and ask again
Do NOT batch the scope question — ask separately for each MCP. Do NOT offer project scope — it modifies .claude/settings.json which is tracked in git.
Commands per MCP:
bashclaude mcp add -s {scope} linear-server --transport http https://mcp.linear.app/mcp
After adding, tell the user to run /mcp in their next session to authenticate.
bashclaude mcp add -s {scope} notion --transport http https://mcp.notion.com/mcp
After adding, tell the user to run /mcp in their next session to authenticate.
Tell the user to run this command themselves (it handles auth via OAuth automatically):
npx ctx7 setup --claudebashclaude mcp add -s {scope} figma --transport http https://mcp.figma.com/mcp
After adding, tell the user to run /mcp in their next session to authenticate.
manual steps remaining (authentication via /mcp, Context7 setup).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,220 | 11,949 | -16% | 1 | 1 | 0% | 2,563 | 995 | -61% | 0 | 0 | — |
case-02 | fail→fail | 16,549 | 4,638 | -72% | 1 | 1 | 0% | 1,939 | 786 | -59% | 0 | 0 | — |
case-03 | fail→fail | 12,000 | 4,657 | -61% | 1 | 1 | 0% | 1,961 | 808 | -59% | 0 | 0 | — |
case-04 | pass→pass | 23,036 | 38,001 | +65% | 1 | 1 | 0% | 4,650 | 5,237 | +13% | 0 | 0 | — |
case-05 | pass→pass | 9,712 | 6,632 | -32% | 1 | 1 | 0% | 1,777 | 1,890 | +6% | 0 | 0 | — |
case-06 | pass→pass | 35,248 | 18,663 | -47% | 1 | 1 | 0% | 3,455 | 4,470 | +29% | 0 | 0 | — |
case-07 | fail→pass | 15,781 | 9,029 | -43% | 1 | 1 | 0% | 1,650 | 1,247 | -24% | 0 | 0 | — |
case-08 | fail→fail | 13,742 | 8,776 | -36% | 1 | 1 | 0% | 1,311 | 1,738 | +33% | 0 | 0 | — |
case-09 | fail→pass | 13,845 | 8,124 | -41% | 1 | 1 | 0% | 1,510 | 976 | -35% | 0 | 0 | — |
case-10 | fail→pass | 12,986 | 8,227 | -37% | 1 | 1 | 0% | 1,194 | 1,001 | -16% | 0 | 0 | — |
case-11 | fail→fail | 24,386 | 46,285 | +90% | 1 | 1 | 0% | 1,860 | 1,238 | -33% | 0 | 0 | — |
case-12 | fail→pass | 22,894 | 2,534 | -89% | 1 | 1 | 0% | 3,385 | 948 | -72% | 0 | 0 | — |
case-13 | fail→fail | 15,751 | 4,360 | -72% | 1 | 1 | 0% | 1,582 | 1,306 | -17% | 0 | 0 | — |
case-14 | fail→pass | 14,943 | 4,182 | -72% | 1 | 1 | 0% | 1,503 | 1,258 | -16% | 0 | 0 | — |
case-15 | fail→pass | 6,473 | 3,675 | -43% | 1 | 1 | 0% | 962 | 1,005 | +4% | 0 | 0 | — |
case-16 | fail→pass | 7,453 | 5,787 | -22% | 1 | 1 | 0% | 1,193 | 886 | -26% | 0 | 0 | — |
case-17 | fail→pass | 14,511 | 4,023 | -72% | 1 | 1 | 0% | 2,448 | 1,296 | -47% | 0 | 0 | — |
case-18 | pass→pass | 12,359 | 3,610 | -71% | 1 | 1 | 0% | 1,994 | 1,051 | -47% | 0 | 0 | — |
case-19 | fail→pass | 9,999 | 6,163 | -38% | 1 | 1 | 0% | 1,599 | 1,553 | -3% | 0 | 0 | — |
case-20 | fail→pass | 9,000 | 3,181 | -65% | 1 | 1 | 0% | 1,406 | 887 | -37% | 0 | 0 | — |
case-21 | fail→pass | 10,441 | 2,268 | -78% | 1 | 1 | 0% | 1,768 | 856 | -52% | 0 | 0 | — |
case-22 | pass→pass | 9,955 | 2,270 | -77% | 1 | 1 | 0% | 1,438 | 886 | -38% | 0 | 0 | — |
case-23 | fail→pass | 12,707 | 8,705 | -31% | 1 | 1 | 0% | 1,915 | 1,975 | +3% | 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 +52 percentage points is the difference between those two pass rates over the 21 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.