Upload documents
- In the web app, open Knowledge, then Documents, and upload files.
- Drag files into the upload area, or click to browse.
- Choose who can access each document. The upload starts only after you save these permissions.
- Wait until the document shows Ready, then ask a question about it.
| Limit | Value |
|---|---|
| File types | PDF (.pdf), Word (.docx), Markdown (.md), plain text (.txt) |
| Size | 5 MB per file |
| Files per upload | 10 |
| File name | 255 characters or fewer |
When a document changes, upload the new version in place of the old one. Answers use the current version, and earlier versions stay listed. Before you delete a document, askneutron shows which agents use it.
Review and expiry dates
When you upload a document, you can set two optional dates:
- Review on: when someone should check the document is still right.
- Expires on: when the document stops being current. It must be on or after the review date.
Organization admins get a reminder 14, 7, and 3 days before each date, and on the day. Reminders arrive in the web app and by email, and in Slack when Slack is connected. Each admin chooses their channels in notification settings.
When an answer uses a document that is past its review or expiry date, askneutron flags it for review in analytics. The age of an upload alone never marks a document as out of date.
Crawl a website
Crawl a help center, docs site, or intranet and keep it current on a schedule.
- Open Knowledge, then Sites, and add a site.
- Give it a label and a start URL, such as
https://docs.example.com/. - Set the crawl rules, access groups, and a sign-in if the pages are private.
- Run a test crawl. It checks that askneutron can reach the site without adding pages to knowledge.
- Start the full crawl.
Crawl rules
| Rule | What it does | Default |
|---|---|---|
| Frequency | Manual, daily, weekly, or monthly. | Manual |
| Discovery | Sitemap and links, sitemap only, or page links only. | Sitemap and links |
| Page limit | The most pages one crawl adds. | 100 |
| Depth | How many links away from the start URL to follow. | 3 |
| Render JavaScript | Load each page in a browser first, for sites that build pages with JavaScript. | On |
| Include subdomains | Also crawl subdomains of the start URL. | Off |
| Include external links | Follow links to other websites. | Off |
| Include and exclude patterns | URL patterns to keep or skip, such as https://docs.example.com/archive/**. | None |
Private pages
To crawl pages behind a sign-in, choose one of these when you add the site:
- Basic auth (username and password)
- Bearer or API token
- Custom headers
- Cookie header
Saved passwords, tokens, and cookies are never shown again. Leave the field blank when you edit the site to keep the saved value.
Connect a tool
Connectors sync content from the tools your team already uses, along with each tool’s sharing rules. Open Integrations in the web app, choose a tool, connect it, and test the connection before you save.
These connectors are in preview:
- Google DrivePreview
- Google SheetsPreview
- SlackPreview
- JiraPreview
- LinearPreview
See every source, including ones coming soon, in the integrations directory.
Who can see what
Every source has an access policy. For uploads and sites, you choose the groups that can use it. Connected tools bring their own sharing rules. When someone asks a question, Neutron only uses sources that person is allowed to see, in the web app, the widget, MCP clients, and the API.