guru-mcp_skill

Overview

This skill integrates Guru via the official MCP server so you can ask AI questions, search your knowledge base, create drafts, and update cards from one interface. It connects to all Guru sources (Slack, Drive, Confluence, SharePoint, Salesforce) and respects workspace permissions. Use it to surface answers, manage card content, and log queries into Guru’s AI Agent Center.

How this skill works

The skill calls Guru MCP endpoints to list Knowledge Agents, generate AI answers, search documents, retrieve card HTML, create drafts, and update existing cards. It requires a Guru API token (email:token) configured in your environment and an mcporter MCP server entry. Responses include content and source metadata so you can verify and preserve original HTML structure when updating.

When to use it

• Get concise, sourced answers to operational questions using AI-driven Knowledge Agents
• Search across all connected sources for existing documentation or cards
• Retrieve full card HTML when you need authoritative content or to copy structure
• Create new draft cards from AI-generated content or templates
• Update or maintain existing cards while preserving DOM and permission constraints

Best practices

• Always call guru_list_knowledge_agents first to obtain the correct agentId
• Restrict queries with collectionIds or sourceIds to reduce noise and improve relevance
• When updating, fetch current HTML with guru_get_card_by_id and preserve the DOM hierarchy—insert or replace within existing structure
• Store GURU_API_TOKEN in a secure environment file and use mcporter configuration for server headers
• Review AI-generated drafts before publishing; log and monitor queries in AI Agent Center for auditability

Example use cases

• Ask a Knowledge Agent: "How do I submit expenses?" and get a sourced, step-by-step answer
• Find related documentation: search for "onboarding process" across Slack, Drive, and Confluence
• Create a new guide: generate an API Authentication Guide draft and get a draft ID/URL for review
• Update a policy card: retrieve card HTML, apply content changes, and push updates preserving HTML structure
• Audit usage: review AI Agent Center analytics to see what questions are asked and which cards are referenced