application/src/Fediverse/RobotsTxt/fetchRobotsTxt.ts

@@ -1,6 +1,5 @@
 import axios, { AxiosRequestConfig, AxiosResponse } from 'axios'
 import robotsParser from 'robots-parser'
-import { getDefaultTimeoutMilliseconds } from '../getDefaultTimeoutMilliseconds.js'
 import RobotsTxt from './RobotsTxt.js'
 import { RobotsTxtError } from './RobotsTxtError.js'
 
@@ -11,10 +10,7 @@ export default async function fetchRobotsTxt (domain: string): Promise<RobotsTxt
   const url = `https://${domain}/robots.txt`
   let content = ''
   try {
-    const robotsTxt = await axios.get(url, {
-      headers: { 'User-Agent': userAgent },
-      timeout: getDefaultTimeoutMilliseconds()
-    })
+    const robotsTxt = await axios.get(url)
     content = robotsTxt.data
   } catch (error) {
     console.info('Robots.txt not found', { error, url })

application/src/Jobs/Feeds/refreshFeedsOnPage.ts

@@ -13,7 +13,7 @@ export const refreshFeedsOnPage = async (
   robotsTxt: RobotsTxt
 ): Promise<Feed[]> => {
   const feedData = await provider.retrieveFeeds(node.domain, page, robotsTxt)
-  const indexableFeedData = feedData.filter(item => item.indexable && !item.description.includes('#nobot'))
+  const indexableFeedData = feedData.filter(item => item.indexable && !item.description.includes('#noindex'))
   console.info('Retrieved feeds', {
     count: feedData.length,
     indexableCount: indexableFeedData.length,