fix: Allow task runner to invoke skills via slash command prefix (#29302)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.claude/plugins/n8n/skills/linear-issue/SKILL.md

@@ -1,6 +1,5 @@
 ---
 description: Fetch and analyze Linear issue with all related context. Use when starting work on a Linear ticket, analyzing issues, or gathering context about a Linear issue.
-disable-model-invocation: true
 argument-hint: "[issue-id]"
 compatibility:
   requires:

.github/scripts/claude-task/prepare-claude-prompt.mjs

@@ -32,6 +32,13 @@ let prompt;
 
 if (useRaw) {
 	prompt = task;
+} else if (task.startsWith('/')) {
+	// Task is a skill invocation (e.g. "/n8n:linear-issue CAT-2820").
+	// Wrap it so the model invokes the Skill tool instead of implementing code.
+	prompt = `# Skill Invocation
+Invoke the following skill using the Skill tool and follow its instructions.
+
+${task}`;
 } else {
 	// List available templates so Claude knows what exists (reads them if needed)
 	const templateDir = '.github/claude-templates';

.github/workflows/util-claude-task.yml

@@ -131,6 +131,8 @@ jobs:
           prompt: ${{ env.CLAUDE_PROMPT }}
           show_full_output: ${{ inputs.suppress_output != true }}
           display_report: false
+          plugin_marketplaces: "./.claude/plugins/n8n"
+          plugins: "n8n@n8n"
           settings: |
             {
               "permissions": {