PrompTugal.news· Data
Version v0.1.151data read 2026-11-21 09:20 UTCPay for what you read · free here

Web Summit 2026 replays on YouTube

The recorded talks of Web Summit 2026 (Lisbon, 9 to 12 November 2026; Day Two, 11 November, cancelled) that Web Summit's official YouTube channel published since 1 November 2026: title, time published, length, view and like counts as read, a link to YouTube, the address for YouTube's own player, and the chapters (talk start times and titles, speakers where the line names them) read from each description's timestamp lines. The videos stay Web Summit's: we link and embed with YouTube's player, and keep no description, thumbnail or comment.

Three layers: the source (who publishes it, the terms, when it was read), the raw data as the publisher served it, and the processed data, what this site makes of it. The processing is where the value is: named fields, places joined, gaps marked, every value kept with its source and time.

1. The source

At a glance

FieldValue
PublisherWeb Summit, through YouTube (YouTube LLC, Google)
Sourcehttps://www.youtube.com/@WebSummit
Reader pagehttps://www.youtube.com/@WebSummit
Scheduleevery 6 hours: the Pages workflow (.github/workflows/pages.yml) at 17 minutes past 00, 06, 12 and 18 UTC; 3 or 4 requests a run (as many quota units, of the 10,000 a day by default; the run stops at 200 units); where the environment has no key (GitHub Actions), the run does nothing and keeps the last read, copied from the live site, until it is 25 days old, then deletes the videos from it
Ownerdata.b@PrompTugal.dinis
Collectorsources/youtube-replays/collect.mjs
Since2026-11-19
Last read2026-11-21 07:13 UTC
Last runok: 3 requests ok, 0 failed
Now6 videos, 0 chapters
LicenceThe data: YouTube API Services Terms of Service and Developer Policies (Web Summit (the videos), YouTube (the API data)). This page, the card and the records we add: CC BY 4.0.

Reuse terms

YouTube API Services Terms of Service and Developer Policies (both last updated 2026-11-06 UTC). Data read without a user's sign-in is Non-Authorized Data: stored for no longer than 30 calendar days, then deleted or refreshed; so this source keeps only the latest read, replaced every run, with no raw copy or history in a vault. YouTube must be shown as the source and its Terms linked (https://www.youtube.com/t/terms); no copy or offline use of the audiovisual content (we link and use YouTube's embedded player only); no aggregation of API Data beyond one content owner's own channels (we read one channel and add no totals across channels). The videos, titles and descriptions stay Web Summit's. Not reviewed by a person. Reuse: per item: the videos, titles and descriptions stay Web Summit's; we publish the link, the embed address and values read through the API, replaced every run (at most 30 calendar days old), with YouTube named as the source and its Terms linked. Commercial use of what YouTube serves is not ours to grant: unclear. A site that sells subscriptions should link and embed the videos with YouTube's player and not charge to watch them..

API Clients may temporarily store limited amounts of Non-Authorized Data for as long as is necessary for the purposes of the API Client but not longer than 30 calendar days. As in section (III.E.4.c) immediately above, this means that after 30 calendar days, the API Client must either delete or refresh the stored data.

Open question: Dinis: (1) who accepted the YouTube API Services Terms of Service for the project the key belongs to (an agent never does)? (2) The Developer Policies ask an API Client to display a link to YouTube's Terms and the YouTube Brand Features where it shows YouTube content: the sites that show these videos (library.b@'s pages, the Website) must do so, and library.b@ should be told. (3) Whether our reading of the 30-day rule is right (latest read only, refreshed every 6 hours, deleted if 25 days old without a refresh) is for a person to confirm. (4) No dated copy of a read is kept anywhere, so there is no way to show later what a count was on a given day; history needs YouTube's written approval or the channel owner's own authorisation.

2. The raw data

There is no raw layer: YouTube's terms let us store API data for 30 calendar days at most, so this source keeps only its latest read, replaced every run, and no raw answer or history anywhere (not in a vault either). Web Summit's channel. Source: YouTube. By using these links and players you are bound by the YouTube Terms of Service.

3. The processed data

The collector asks the YouTube Data API for the channel Web Summit (@WebSummit), its uploads playlist back to 1 November 2026 and each video's details (3 or 4 requests, as many quota units, never search), and keeps each video's id, title, time published, length, view and like counts with the time read, a link to YouTube and the address of YouTube's embedded player, and chapters read from timestamp lines in the description. The description, thumbnails and comments are not kept. The key is added by the environment, never held by the code; with no key the run does nothing and keeps the last read, and a read more than 25 days old is deleted. Content: YouTube; the videos stay Web Summit's. The result: latest.json.

Web Summit 2026 replays: each upload with its time, length, counts as read and chapters, with a link to YouTube. The data: /api/v1/youtube-replays/latest.json.

More about this source

What runs by code, by agents and by people

Code

Agents

People

Cost

FieldValue
per run tokens0
per run model cost usd0
runnerGitHub Actions on a public repository: no charge; 3 or 4 requests and as many quota units a run (cap 200)
build tokensnot measured
noteCode only: no model is called when the data is collected.

API addresses

These addresses stay for the life of v1 (the contract).

FileAddress
Latest readhttps://data.promptugal.news/api/v1/youtube-replays/latest.json
Every readhttps://data.promptugal.news/api/v1/youtube-replays/reads.json
Cardhttps://data.promptugal.news/api/v1/youtube-replays/card.json
OpenAPIhttps://data.promptugal.news/api/v1/youtube-replays/openapi.json
curl -s https://data.promptugal.news/api/v1/youtube-replays/latest.json
curl -s https://data.promptugal.news/api/v1/youtube-replays/reads.json | head -c 600

Vault

Vault not yet created. The collector writes the bytes of each read to a vault copy; until the vault exists they are not kept.

Last runs

Time (UTC)Requests okRequests failedReads with changed bytes
2026-11-21 07:13 UTC301 of 3
2026-11-20 18:16 UTC302 of 3
2026-11-20 07:17 UTC301 of 3
2026-11-19 13:08 UTC301 of 3
2026-11-19 07:07 UTC301 of 3
2026-11-19 01:28 UTC303 of 3
2026-11-19 01:20 UTC010 of 1
2026-11-19 01:16 UTC010 of 1
2026-11-19 00:59 UTC010 of 1
2026-11-19 00:29 UTC011 of 1

Showing 10 of 10 runs kept. Full detail with sha256: reads.json.

Preview of the values

The first 2 of 2 entries in library_title_checks, as read 2026-11-21 07:13 UTC. Source: https://www.youtube.com/@WebSummit. Long values are shortened here; the file has them whole.

idexpected_titleread_titlematchSource
S9RHQGNIaFYWeb Summit 2026 | Day OneWeb Summit 2026 | Day Onetruesource
dJi_pXLxwnoWeb Summit 2026 | Opening NightWeb Summit 2026 | Opening Nighttruesource