# Train Atlas on your knowledge base

> Set up sources, keep answers accurate, and securely connect Claude Code or Codex to search and manage your TicketCord knowledge base.

## What Atlas can ingest

You add sources from the dashboard’s knowledge-base config. There are three:

- **Single page.** Point Atlas at a URL and it will crawl the page (and a few configurable levels deeper). Best for help docs, FAQs, and product pages you already maintain on your site.
- **Discord channel.** Import the message history from a Discord channel - useful when your support team has already answered the same questions in public, and you want Atlas to learn from those answers.
- **GitHub repository.** Link a repo and an optional path; Atlas indexes every Markdown file (`.md`, `.mdx`) it finds there. Good for engineering teams who keep their docs in the same place as their code.

## Structuring entries

Smaller, focused sources work better than one giant page. A help center with a separate article per topic (“Refund policy”, “Cancellations”, “Region availability”) is much easier for Atlas to pull clean answers from than a single “Everything you need to know” mega-doc. Atlas cites the source it used in its reply, so granular sources also make the citations more useful.

## Letting staff teach inline

When staff correct an Atlas draft before sending, the correction is captured as a Q&A entry and added to the knowledge base. It then competes with your other sources on the next similar question - ranked by relevance and by community votes - so good corrections rise on their own over time. No retraining job to run; new entries are searchable immediately.

## What not to put in

- Secrets, passwords, API keys, customer data.
- Outdated policies. Atlas treats every entry as current.
- Long-running incident notes. Use canned responses for those instead.

## Quality checks

Open the knowledge base list in the dashboard every so often and skim it for outdated entries. If an answer changed (pricing, policy, supported region), edit or remove the entry so Atlas stops citing the stale version. Re-crawl any URL-based entries after big docs changes so the index reflects what your customers will actually find on your site.

## Manage your knowledge base with AI tools

**MCP** is a secure bridge that lets Claude Code or Codex work with your TicketCord knowledge base after you sign in and approve access. You can ask in plain language instead of clicking through every dashboard screen.

## Before you connect

- You must be the TicketCord account owner. Dashboard staff cannot create an MCP connection.
- Your live plan must include Knowledge Base access.
- Install Claude Code or the Codex CLI, and make sure you can sign in to TicketCord.
- One connection can reach every server you own now or add later. Each request still names one bot and server, and TicketCord rechecks ownership and plan access every time.

## Connect Claude Code or Codex

1. In the dashboard, select a bot and server. Open **Knowledge Base > Configuration**, then find **Connect AI tools**.
2. For Claude Code, run `claude mcp add --transport http ticketcord https://ticketcord.com/api/mcp --scope user`, then `claude mcp login ticketcord`.
3. For Codex, run `codex mcp add ticketcord --url https://ticketcord.com/api/mcp`, then `codex mcp login ticketcord`.
4. Your browser opens TicketCord. Sign in, read the account-wide warning and the requested read, write, and delete permissions, then choose **Allow access** or **Deny**.
5. When you see **Authentication complete**, close the tab and return to your AI tool. Try: `List my TicketCord knowledge bases.`

## What you can ask it to do

- List your knowledge bases and inspect settings, sources, indexing status, and entries.
- Search by meaning and return short matching passages with source labels and confidence.
- Add, edit, enable, disable, reindex, or delete webpage, Discord channel, and already-connected GitHub sources.
- Create and edit entries, review pending entries, and approve, reject, or delete them.
- Update dashboard-editable settings for product details, triggers, display, quality, updates, behaviour, feedback, ticket learning, and Auto Support.

## Useful requests to try

- `List my knowledge bases and show which ones have no ready sources.`
- `Search our refund policy and summarize it with source labels.`
- `Find entries that mention old pricing. Show them first and do not change anything.`
- `Add https://docs.example.com as a documentation source and start indexing it.`
- `Show pending entries and recommend approve or reject before making changes.`
- `Show me the current password-reset answer, then help me update it.`

## Permissions, safety, and access

- Read access can view and search knowledge. Write access can change settings, sources, and entries. Delete access can permanently remove sources and entries.
- For risky changes, ask the AI tool to show what it found and explain its planned action before it changes anything.
- Deleting is permanent. TicketCord checks your permission again for every request, even after you approved the connection.
- If your plan expires or loses Knowledge Base access, the connection pauses. It can work again after an eligible upgrade unless you revoked it or its refresh access expired.
- Revoke a connection immediately from the **Connect AI tools** card in the Knowledge Base configuration.
- Connecting or disconnecting a GitHub account still happens only in the dashboard. Never store passwords, API keys, secrets, or customer data in your knowledge base.

## Machine-readable resources

- [Canonical HTML](https://ticketcord.com/help/ai/knowledge-base)
- [TicketCord llms.txt](https://ticketcord.com/llms.txt)
- [AI page index](https://ticketcord.com/api/ai/pages)