# kapa.ai documentation

> Documentation for kapa.ai, a platform that indexes a company's technical knowledge (documentation, source code, GitHub issues, support tickets, community forums, PDFs, and 50+ other sources) into a unified knowledge base and uses a retrieval-augmented generation (RAG) answer engine to answer technical questions with inline citations. These docs cover connecting data sources, deploying Kapa across surfaces (widget, Slack/Discord bots, support deflector, MCP server, SDKs), customization, analytics, and the HTTP API.

Key facts for interpreting these docs:

- This is the product documentation host (https://docs.kapa.ai). The marketing site is https://www.kapa.ai.
- Kapa is model-agnostic and optimized for accuracy: answers are grounded in the customer's connected sources, and Kapa says "I don't know" rather than hallucinating when sources do not cover a question.
- The Kapa platform is at https://app.kapa.ai. The public HTTP API base URL is https://api.kapa.ai (authenticated with an `X-API-KEY` header).
- Kapa hosts a public documentation MCP server at https://kapa-docs.mcp.kapa.ai (Streamable HTTP), exposing the tool `search_kapa_knowledge_sources` for use in Claude, Cursor, ChatGPT, and other agents.

## Getting started

- [Overview](https://docs.kapa.ai/): What Kapa is and the main ways to deploy it.
- [Index your first source](https://docs.kapa.ai/getting-started/index-your-first-source): Connect your sources and let ingestion build the knowledge base.
- [Connect an AI agent](https://docs.kapa.ai/getting-started/connect-an-ai-agent): Hand an agent search over your knowledge as an MCP tool.
- [Build with AI](https://docs.kapa.ai/build-with-ai): Use Kapa's MCP server in AI coding tools to implement Kapa correctly.
- [FAQ](https://docs.kapa.ai/retrieval/faq): What problem Kapa solves, models used, language support, multi-instance setup.

## Connecting knowledge sources

- [Knowledge sources](https://docs.kapa.ai/knowledge-sources/): Supported connectors (web crawl, GitHub, Slack, Zendesk, Confluence, Notion, OpenAPI, and more).
- [Data ingestion](https://docs.kapa.ai/knowledge-sources/data-ingestion): How ingestion, chunking, and automatic refresh work.

## Deploying Kapa

- [All integrations](https://docs.kapa.ai/integrations/): Every supported way to deploy Kapa.
- [Website widget](https://docs.kapa.ai/integrations/website-widget): Embed an "Ask AI" chat widget with a single script tag.
- [Support form deflector](https://docs.kapa.ai/integrations/support-form-deflector): Suggest answers before a support ticket is submitted.
- [MCP server](https://docs.kapa.ai/retrieval/hosted-mcp-server): Deploy a hosted Model Context Protocol server for your knowledge sources.
- [In-product agents](https://docs.kapa.ai/use-cases/in-product-agents): Combine Kapa retrieval with your own tools inside your product.
- [Slack bot](https://docs.kapa.ai/integrations/slack-bot): Answer questions inside a Slack community or workspace.
- [Discord bot](https://docs.kapa.ai/integrations/discord-bot): Answer questions inside a Discord community.
- [Zendesk agent](https://docs.kapa.ai/integrations/zendesk-agent): Assist support agents inside Zendesk.
- [Internal technical assistant](https://docs.kapa.ai/integrations/internal-technical-assistant): An internal assistant for support, CS, and solutions teams.

## Customizing

- [Customizing Kapa](https://docs.kapa.ai/customizing): Adjust tone, scope, and behavior via system-instruction customizations.

## Analytics

- [Analytics overview](https://docs.kapa.ai/analytics/): Understand what users ask and how the assistant performs.
- [Coverage gaps](https://docs.kapa.ai/analytics/coverage-gaps): Find where documentation is missing or insufficient.
- [Top questions](https://docs.kapa.ai/analytics/top-questions): The most frequent questions users ask.

## Developer tools

- [API overview](https://docs.kapa.ai/retrieval/http-api/): HTTP API for custom integrations.
- [API reference](https://docs.kapa.ai/api/reference): Full endpoint reference (base URL https://api.kapa.ai).
- [Integrations FAQ](https://docs.kapa.ai/integrations/faq): API keys, integration IDs, and common questions.
- [Chat SDK](https://docs.kapa.ai/dev/sdk/): React/JS SDK for building fully custom front-ends.
- [Agent SDK](https://docs.kapa.ai/dev/agent): Build in-product agents with streaming, custom tools, and a chat UI.

## Skills

- [Kapa Skills](https://docs.kapa.ai/skills): AI agent skills for documentation workflows in Claude Code, Cursor, and Codex. Available skills:
  - Analyze Coverage Gaps: Turn your Coverage Gaps export into per-cluster files and work through each with your agent to write the missing content directly in your repo.
  - Analyze Source Analytics: Audit the pages Kapa cites most often against the questions they were cited in, using a subagent per page, then fix the docs on the spot.
  - Analyze Top Questions: Turn your Top Questions export into topic files and refine existing docs to match how users actually ask.
  - Answer RFPs: Extract every requirement from an RFP or security questionnaire, retrieve answers live from your knowledge base via the Hosted MCP server, and generate the final response (Word, PDF, or Excel).

## Security and account

- [PII masking](https://docs.kapa.ai/security/pii-masking): How Kapa removes PII from user messages and knowledge sources.
- [Security certifications (SOC 2)](https://docs.kapa.ai/security/certifications): Compliance and certification details.
- [Account management](https://docs.kapa.ai/account/): Projects, roles and permissions, and billing.

## Optional

- [Improving Kapa](https://docs.kapa.ai/improving/writing-best-practices): Documentation best practices to improve answer accuracy.