🗨️ PackIA: AI Chatbots & Assistants
Assistants powered by your knowledge, without hallucinations
Direct answer: Assistants powered by your knowledge, without hallucinations This pack brings together 7 resources for Claude, Claude Code, ChatGPT Codex.
RAG chatbots that answer from YOUR documents: ingestion pipelines, system prompts with guardrails, quality evaluation, and deployment on web, WhatsApp, or Slack.
Included in the Completo — $50
Plans are cumulative: the Pro Plan includes all Essential Plan packs, and the Complete Plan includes all packs (Essential + Pro). This pack unlocks with the Completo ($50).
Buy with the Completo ($50) Or build your custom plan
For Claude: the skills, MCP servers, prompts, and guides in this pack are used in conversation with Claude, which consults them while working with your files and connected services. View for Claude: only the resource types compatible with Claude (Skill, MCP Server, Prompt collection, Guide). The full pack has 7 resources.
Use cases
- RAG chatbot trained on your documentation in hours
- Guardrails: what to answer, what to escalate, what to never say
- Evaluation with 50 golden questions before launch
Master prompt: Chatbot Engineer
You are a chatbot engineer. Case: {case}, knowledge: {sources}, channel: {channel}, users: {users}. 1) RAG pipeline: chunking, embeddings, and retrieval with re-ranking, 2) system prompt: role, tone, 'I don't know' policy, escalation to a human, and banned topics, 3) guardrails: input validation, length limits, prompt injection protection, 4) test set of 50 questions with expected answers and metrics, 5) deployment and continuous monitoring plan.
Included resources (6)
- qdrant (MCP Server, qdrant/mcp-server-qdrant, ★ 1.5k · official) — Vector storage and semantic search: long-term memory for agents. GitHub
- claude-api (Skill, anthropics/skills, ★ 176.6k · official) — Claude API reference: models, pricing, parameters, streaming, tool use, MCP, and token counting. GitHub
- fetch (MCP Server, modelcontextprotocol/servers, ★ 90.4k · official) — Official reference server: fetch web content and convert it to token-efficient markdown for LLMs. GitHub
- tavily (MCP Server, tavily-ai/tavily-mcp, ★ 2.4k · official) — Production-ready real-time web search: extraction, sitemaps, and crawling. GitHub
- mcp-builder (Skill, anthropics/skills, ★ 176.6k · official) — Official guide for creating high-quality MCP servers in Python (FastMCP) or TypeScript, with tools and evaluation suites. GitHub
- slack (MCP Server, zencoderai/slack-mcp-server, ★ 76 · official) — Manage Slack channels, history, and messaging; the maintained successor to the reference server. GitHub