agentleFS
agiea-brainagiea-brain/agency-skills/skills/hunter-harris/demandbird-exact-time-linkedin-post/SKILL.md

demandbird-exact-time-linkedin-post

Skill1 starsChanged 16 days agoMITView on GitHub

Use this skill when an approved LinkedIn post (text, image or PDF carousel) must go out at a specific minute through DemandBird. Produces a scheduled post whose stored copy, time and media match what was approved, confirmed by one read-back. Uses DemandBird's MCP server. "Schedule this post for Tuesday 10am", "queue the carousel", "schedule the week's LinkedIn posts".

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 2 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Call list_connected_accounts and list_scheduled_posts.

    line 21
  2. Very specific:

    Call schedule_post with the post ID, scheduled_at as ISO 8601 with the UTC offset for that date, and variance_minutes: 0.

    line 24
  3. Very specific:

    Call get_post once. Compare it with the approved file: copy (only a trailing newline may differ), time, media and status.

    line 25
  4. Distinctive:

    MUST have approval for the copy, media, account and dated time of every post, batch rows included.

    line 38
  5. Very specific:

    MUST set variance_minutes: 0 when the time matters.

    line 39
  6. Very specific:

    MUST stop a batch on the first rejection, wait, and reconcile with list_scheduled_posts before retrying.

    line 40
  7. Very specific:

    NEVER mark a first comment verified from get_post.

    line 41
  8. Distinctive:

    NEVER give the agent a token that can delete posts or manage channels.

    line 42

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Applies when approved LinkedIn copy has to land at an exact time, often for a client or a launch. Produces a scheduled DemandBird post that matches the approval, with the read-back saved as proof.

Needs DemandBird's MCP server connected with a token scoped to drafting and scheduling. Leave deletion out of the token.

The play

  1. Call list_connected_accounts and list_scheduled_posts. Take the LinkedIn account ID from the result, never from memory, and check nothing is already queued in that slot.
  2. Save the approved copy verbatim with save_draft: content, platforms: ["linkedin"], the account in platform_account_ids, and any media in images. Don't run it through a drafting or voice tool; that rewrites approved copy.
  3. For a PDF carousel, set the document title explicitly. Otherwise LinkedIn shows the filename.
  4. Call schedule_post with the post ID, scheduled_at as ISO 8601 with the UTC offset for that date, and variance_minutes: 0. Without the zero, the account's default jitter moves the time.
  5. Call get_post once. Compare it with the approved file: copy (only a trailing newline may differ), time, media and status. Save that read-back. Don't poll.
  6. Record the post as scheduled with its ID and time. There is no public URL until it publishes; fill it in afterwards.

Media and first-comment quirks, and the rate-limit behaviour, are in references/demandbird-quirks.md. Open it before a batch.

What good looks like

  • First check: the stored time and variance. A post "scheduled for 3pm" that stores a window instead of 3:00 never had its variance set to zero.
  • The mediocre version loops save_draft over a week of posts. The first rate-limit rejection doesn't stop the loop, and every later call fails or, worse, half-succeeds.
  • The output is right when each read-back matches the approved copy byte for byte (bar the trailing newline), the time is exact in the offset for that date, and there is exactly one post per slot.

Rules

  • MUST have approval for the copy, media, account and dated time of every post, batch rows included; save the copy verbatim.
  • MUST set variance_minutes: 0 when the time matters.
  • MUST stop a batch on the first rejection, wait, and reconcile with list_scheduled_posts before retrying. A rejected write doesn't prove nothing was created.
  • NEVER mark a first comment verified from get_post; it doesn't return them.
  • NEVER give the agent a token that can delete posts or manage channels.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.