Post to YouTube
Upload a video, post on the Posts tab, comment, reply, like, and collect analytics, with the behavior of each activity.
Upload a video to YouTube, post on your channel's Posts tab, comment, reply or like, and collect analytics, from a content activity.
Uploads through YouTube Studio, in signed-in Google Chrome, are the supported path and the default. One hand-supervised private Studio upload has already been run (2026-11-19). Unlisted and Public, Checks, the made-for-kids choice, and many dialog selectors have not been live-clicked; they are modeled from YouTube's pages and docs.
Uploads through the YouTube Data API are experimental and have not been live-verified. Giving the Data API consent switches uploads, comments, replies and likes onto that path; there is no fallback to Studio. Without the consent, uploads use Studio. Analytics also need that consent. Comments, replies, likes and analytics through the Data API have not had a live check either. What each activity does, including that caveat, is also on the YouTube reference page.
Before the first run
-
Let the extension reach YouTube in the Google Chrome profile this Blowhorn profile uses, and sign in to the channel by hand. In the desktop app, open Profiles, open the profile, go to its Browser tab, and choose Turn on YouTube. From Terminal:
terminal blowhorn chrome open-profile your-profile --sign-in youtubeAllow YouTube when the Blowhorn page and Chrome ask, then sign in on the YouTube page it goes on to. Until that permission is granted, a YouTube run is refused with nothing sent. Discover does not look for a YouTube sign-in. The full procedure is in Set up your platform accounts.
-
Store the channel's address by ID, shown at youtube.com/account_advanced. A handle address (
/@name) is refused, because Blowhorn never guesses which channel a handle means.terminal blowhorn profile set your-profile YT_CHANNEL_URL=https://www.youtube.com/channel/UC... -
Optional: give the Data API consent if you want analytics, or if you want comments, replies and likes to go through the YouTube Data API instead of Google Chrome. Uploads through the Data API are experimental and not live-verified; Studio remains the supported default. The consent uses your own Google Cloud OAuth client (type Desktop app); Blowhorn ships none. The steps are in Give the YouTube Data API consent.
terminal blowhorn profile auth --platform youtube --profile your-profileWith that consent in place, uploads, comments, replies and likes use the Data API (uploads on that path are experimental). Without it, those activities use signed-in Google Chrome, which is the supported default for uploads, and analytics skips the profile. A failure on one path never falls back to the other.
If the profile has no YouTube credential, blowhorn post --profile your-profile --platform youtube exits non-zero and says the profile has
no credential for youtube, so nothing can be selected for posting, and
to add one with blowhorn profile set your-profile <KEY>=<value>. That
message does not name YT_CHANNEL_URL. Set the key as shown above. A
handle address is refused by name, with the ID form to use. Nothing
opens Google Chrome in either case.
Upload a video
Set Video URL, Title and Message Text (the description). Leave
Comment on, Amplify and Community blank. A blank Visibility is
private.
blowhorn content upsert --payload '{"Platform":"YouTube","Profile":"your-profile","Title":"Launch day","Message Text":"What we shipped and why it matters.","Video URL":"https://example.com/launch.mp4","Approved?":"yes"}'Then preview, and publish:
blowhorn post --platform youtube --profile your-profile --dry-run
blowhorn post --platform youtube --profile your-profileThrough Google Chrome, Blowhorn opens Studio's upload page at the
channel ID in YT_CHANNEL_URL, attaches the file, and walks Details,
Video elements, Checks and Visibility. It answers "made for kids" with
No, and stops if Studio Checks reports a problem, with nothing
published. The channel is checked when the upload page shows, once
Details opens, and again just before the final click, because Studio can
switch channels after the upload page first showed. If the address is
not the channel YT_CHANNEL_URL names, the activity stops with nothing
published.
Visibility is chosen explicitly and verified before that click. Studio's
Visibility page starts on Public, with the final button reading Publish.
Blowhorn selects the row's setting (private when Visibility is
blank), reads the choice back, and clicks only when the button reads
Save (private or unlisted) or Publish (public). If the button does not
read what that setting requires, the click is refused and nothing is
published. Unlisted and Public have not been live-clicked.
Nothing is read after a successful click, so the activity is recorded as clicked, outcome not read: check the upload in YouTube Studio.
The title may be at most 100 characters and the description at most
5,000. A longer value is refused before the file is fetched and before
Studio opens, not shortened. < and > are refused in both.
Once the file is attached, Studio has already saved the upload as private, including during a dry fire through Google Chrome or a walk that later stops. Delete that draft in YouTube Studio if you do not want it. A retry starts a new upload; it does not resume the leftover draft.
Through the Data API, which is experimental for uploads and has not
been live-verified, the same row is sent as a resumable upload under
your own OAuth client. The file goes in chunks (8 MiB by default, in
256 KiB units). If the connection drops, the next run of the same row
(same file, channel and metadata) resumes from the last byte YouTube
confirmed, never from a byte it did not. After six failures in a row
with no progress (with exponential backoff capped at 64 seconds), the
row is held and the session is kept for the next run. If the last chunk
went out and nothing could be read after it, the activity is recorded as
sent, outcome not read, and is not blindly retried. A quota,
uploadLimitExceeded, rate-limit or 429 answer holds the row instead of
failing it, and stops further Data API work for the rest of the run.
This path has not been live-checked.
On the Data API path no Visibility page is walked. The visibility is
sent with the upload and read back from YouTube's answer; a different
visibility, or none, is reported but does not undo the upload, which
has already happened. YouTube may lock uploads from an API project that
has not passed Google's audit to private, so a public or unlisted
upload says so before it starts. Set
BLOWHORN_YOUTUBE_API_PROJECT_UNVERIFIED=false once your project has
passed the audit to stop that warning.
Each profile may upload 3 videos a day, counted across every run that day. An activity past that ceiling waits for the next day. A dry run counts its planned rows against that ceiling for the rest of the same run, so it holds what the real run would; it does not stamp the activity, so the count does not carry into tomorrow.
Choose visibility
Set Visibility to private (the default), unlisted or public. Any
other value is refused. These settings belong only to YouTube: a
LinkedIn, X or other activity that sets Visibility, Short or
Community is refused by name (leave the setting blank).
blowhorn content upsert --payload '{"Platform":"YouTube","Profile":"your-profile","Title":"Launch day","Message Text":"What we shipped and why it matters.","Video URL":"https://example.com/launch.mp4","Visibility":"unlisted","Approved?":"yes"}'Through Google Chrome, the choice is verified on Studio's Visibility
page before the final click, as described above. Unlisted and Public
have not been live-clicked. Through the Data API, the visibility is only
read back from YouTube's answer after the upload, and a public or
unlisted upload may land private on an unaudited API project, as
described above.
Upload a Short
Set Short to yes. The video must be 3 minutes or less, and vertical
or square (not wider than it is tall). A row that fails either check is
refused once the file has been downloaded, before Studio's dialog
proceeds (or, on the Data API path, before any byte is sent). A
plan-only --dry-run does not download the file, so it does not measure
length or shape and will not catch a Short that is too long or too wide.
blowhorn content upsert --payload '{"Platform":"YouTube","Profile":"your-profile","Title":"Launch in 30 seconds","Message Text":"A quick look at what shipped.","Video URL":"https://example.com/short.mp4","Short":"yes","Approved?":"yes"}'A blank Short means the video is not a Short. Use yes or no; a JSON
true or false is stored as yes or no. The string "true" is
refused.
Post on the Posts tab
Set Community to yes and put the text in Message Text. Optionally
set Image URL to one image. Leave Title, Video URL, Comment on
and Amplify blank: any of those, or an @ tag in the text, is refused
by name. YouTube states no text length, so Blowhorn does not check one.
blowhorn content upsert --payload '{"Platform":"YouTube","Profile":"your-profile","Message Text":"The launch stream starts at 15:00 CT.","Community":"yes","Approved?":"yes"}'A Posts-tab post runs only in signed-in Chrome. There is no Data API call for it. A scheduled run holds the activity for a run you start yourself. Each profile may publish 2 Posts-tab posts a day.
Comment, reply or like
A comment puts the video's address in Comment on and the text in
Message Text. A reply uses the comment's own address, the watch URL
with &lc=. A like puts the video's address in Amplify and leaves
Message Text blank.
blowhorn content upsert --payload '{"Platform":"YouTube","Profile":"your-profile","Comment on":"https://www.youtube.com/watch?v=VIDEO_ID","Message Text":"The 2:15 mark is the part we would start with.","Approved?":"yes"}'blowhorn content upsert --payload '{"Platform":"YouTube","Profile":"your-profile","Comment on":"https://www.youtube.com/watch?v=VIDEO_ID&lc=COMMENT_ID","Message Text":"Agreed; that is the cut we kept.","Approved?":"yes"}'blowhorn content upsert --payload '{"Platform":"YouTube","Profile":"your-profile","Amplify":"https://www.youtube.com/watch?v=VIDEO_ID","Approved?":"yes"}'An @name in a comment or reply is refused before anything is typed. A
reply to a reply is refused. A like that carries text is refused. A
scheduled run holds every comment, reply and like for a run you start
yourself: YouTube treats automated engagement as spam.
If the Like button already reads as pressed (aria-pressed=true),
Blowhorn leaves it alone and records the like as done. That differs from
Facebook, where an already-reacted post is left unclicked and is not
recorded as done.
Each profile may send 10 comments, 10 replies and 20 likes a day. A dry run counts its planned rows against those ceilings for the rest of the same run, so it holds what the real run would; it does not stamp the activity. With a Data API consent these go through the API; without one they go through the signed-in watch page. A Data API quota, rate-limit or 429 answer holds the row and stops further API work for the rest of the run. The project's quota budget is shared by every profile in the run, because the quota belongs to your API project, not to one profile.
Collect analytics
blowhorn analytics --platform youtube --profile your-profileAnalytics are read-only metrics, collected through the profile's own Data API consent (your own OAuth client). A profile without a consent is skipped. Blowhorn never writes or changes figures on YouTube.
Figures are kept for 30 days, as YouTube API Services Developer Policies III.E.4 require. A collection deletes expired YouTube readings first, before it reads anything. The scheduler also does this on its first real pass of each calendar day, across every profile in the organization, so figures do not outlive 30 days when nobody collects.
If you take the consent back, only that profile's stored YouTube figures are deleted. Other profiles' figures stay. How to take the consent back, and what happens if the store is down, is in Give the YouTube Data API consent. The steps and the history commands are in Collect and read your analytics.
Preview without publishing
A plan-only dry run validates the row and prints the plan. It never opens Studio or sends to the API:
blowhorn post --platform youtube --profile your-profile --dry-runWithout a Data API consent, a dry fire (blowhorn post --dry-run --payload-stdin) walks Studio's dialog as far as Visibility, chooses the
row's setting, reads the final button back, and does not click it. Studio
has already saved a private draft once the file was attached; delete it in
YouTube Studio afterward. With a consent, the same dry fire goes through
the Data API and is plan-only: it prints the plan, opens no Studio page,
sends nothing and leaves no draft.
cat youtube-order.json | blowhorn post --dry-run --payload-stdin --jsonA dry run still counts planned YouTube rows against the profile's daily ceilings for the rest of that run, as described above. It does not spend the Data API project's unit budget, because nothing is sent.
Related
- YouTube - what each activity does, the Data API consent, ceilings and limits.
- Set up your platform accounts - the sign-in and the Data API consent, step by step.
- Preview and publish queued posts - the posting run for every platform.
- Content queue columns and activity states - Visibility, Short and Community.
- Collect and read your analytics - history and the HTML report.