Content Engine Official

Marketingv1.0.016 components

Founder-led content for a founder or a small team. Every morning the desk listens (Reddit, LinkedIn, X, YouTube) for what your buyers are asking, turns the best of it into ideas by pillar, and drafts posts in your voice for LinkedIn and X. You approve; the desk hands you the final text to paste and post. Comments and reactions on what you published are harvested, and the people who match your ICP become leads. A newsletter issue is assembled from the week. Monday tells you which pillar and format earned attention. Comes with Theo, a Content Producer.

Installs in about 20 min, verified with a smoke test.

Install this pack How the install works

How the pack is wired

One named coworker, the agents it orchestrates, and the data, knowledge, and tools underneath. This is the real manifest, not a mockup.

AI EmployeeTheo, Content Producer coordinator
AgentsListenerRepurposerDrafterRewriterPost PrepEngagement HarvesterNewsletter AssemblerWeekly Content Digest
Data & KnowledgeIdeas read/writePosts read/writeListening read/writeEngagements read/writeNewsletter Issues read/writeRepurpose read/writeVoice and Pillars RAG
ToolsScrapecreatorsMonid Apify LinkedinSerperMailchimpTurtle Notify

The employee routes each request to the right worker. Agents read and write the tables, ground answers in the knowledge bases, and act through your connected tools. Every write is gated and traced.

The full blueprint

Every agent's tasks, every table's schema, every trigger. Read it all before you sign in.

Theo, Content Producer

Curious, disciplined, allergic to hashtags. Theo keeps the calendar full, the voice consistent, and never posts a word the founder has not approved. He posts nothing himself; he hands over the text.

8 agents · 6 tables · 1 KBs
Listener
Researcher who reports what people actually said, with links
2 tasks

Searches Reddit, LinkedIn posts, X and YouTube for the playbook's queries, writes the best items as Listening rows and turns the top three into Ideas.

ListeningIdeasScrapecreatorsMonid Apify LinkedinSerper
  1. 01Read the listening queries and pillars in Voice and Pillars. Search each query on Reddit, LinkedIn posts, X and YouTube for items from the last 7 days. Check the Listening table so you do not repeat a URL. Keep at most eight items with real engagement or a clear pain statement. Output: The items found, each with channel, url, a one-line summary and why it matters, plain text.
  2. 02Create a Listening row per item (Item, Channel, Query, URL, Summary, Why It Matters, Found today, Status New). Pick the best three and create an Ideas row for each (Idea, Pillar, Angle, Source Listening, Evidence URL, Status New). Write a three-line summary. Plain text. Output: The three-line summary, plain text.
Repurposer
Editor who finds the five posts hiding in one document
2 tasks · 3 in

Reads a blog post, a doc or a call transcript and writes five ideas across the pillars, each with an angle.

IdeasRepurpose SourcesSerper
  1. 01Source "{{source}}" ({{kind}}) at {{url}}. Read it. Find five ideas across different pillars, each with an angle and the supporting line. Output: Five ideas with pillar, angle and the supporting line, plain text.
  2. 02Create five Ideas rows (Idea, Pillar, Angle, Source per the kind, Evidence URL, Status New). Set the Repurpose row's Status to Done. Write a one-line summary. Plain text. Output: The one-line summary, plain text.
Drafter
Ghostwriter who sounds like the founder, not like a ghostwriter
2 tasks · 5 in

Writes a LinkedIn post and an X variant for a picked idea, hook first, in the voice rules, as Posts rows for approval.

IdeasPosts
  1. 01Idea "{{idea}}", pillar {{pillar}}, angle {{angle}}, source {{source}}, evidence {{evidence_url}}. Read the voice rules and the ten posts that worked. Write the LinkedIn post (under 1,300 characters, line breaks between thoughts) and the X variant. Output: The LinkedIn post and the X variant, plain text.
  2. 02Create two Posts rows: Post ("<idea> (LinkedIn)" and "<idea> (X)"), Idea, Pillar, Channel, Format (Post; Thread when X needs it), Draft, Status Drafted, Publish At (the next weekday for LinkedIn, the day after for X). Set the idea's Status to Drafted. Write a one-line summary. Output: The one-line summary, plain text.
Rewriter
Editor with a red pen
2 tasks · 5 in

Cuts a draft to its point in the voice rules and puts the new version in the Draft column for another look.

Posts
  1. 01Post "{{post}}" for {{channel}} ({{format}}), pillar {{pillar}}. Draft: {{draft}} Rewrite it shorter and sharper in the voice rules; keep the idea. Output: The rewritten draft, plain text.
  2. 02Write the new version into the post's Draft, set Status to Drafted, and say in one line what you cut.Output: One line, plain text.
Post Prep
Producer who makes posting a paste
2 tasks · 6 in

Formats an approved post for its channel, sets it Ready to post, and emails the founder the text to paste with the link to the row. The founder posts by hand and pastes the URL back.

PostsTurtle Notify
  1. 01Post "{{post}}" for {{channel}} ({{format}}), planned {{publish_at}}, status {{status}}. Draft: {{draft}} Format it for the channel per the voice rules. Do not change the words beyond formatting and the length limit. Output: The final text, plain text.
  2. 02Write the final text into Final Text, set Status to Ready to post, and email the founder with Turtle Notify: channel, planned time, the final text, and "paste the URL back into the row once posted". Confirm in one line. Output: One line, plain text.
Engagement Harvester
Researcher who reads the comments so the founder can reply to the right ones
2 tasks · 4 in

Reads the published post's comments and reactions, updates the counts, and creates Engagements rows with the people who match the ICP flagged.

PostsEngagementsScrapecreators
  1. 01Post "{{post}}" on {{channel}} at {{url}}, published {{publish_at}}. If the URL is blank, say so and stop. Read the post's reactions and comments. Check the Engagements table so you do not repeat a person on this post. Output: The counts and the new commenters and notable reactors, plain text.
  2. 02Update the post's Impressions, Reactions and Comments when the source gives them. Create an Engagements row per new person (Engagement "Comment on <post> by <who>", Post, Who, Title, LinkedIn, What They Said, ICP Match, Status New). Write a one-line summary. Output: The one-line summary, plain text.
Newsletter Assembler
Editor who ships the newsletter on time
2 tasks · 3 in

Writes the issue (three sections, one link each) from the posts published this week and creates the draft campaign in the email tool for review.

PostsNewsletter IssuesMailchimp
  1. 01Issue "{{issue}}" for {{date}}, theme: {{theme}}. Read the Posts with Status Published in the last 7 days and the voice rules. Write the issue: three sections, one link each, under 400 words. Output: The issue text, plain text.
  2. 02Create a draft campaign in Mailchimp with the issue (subject: the theme). Do not send. Write the draft link into Draft Link, keep Status Drafted, and confirm in one line. Output: One line, plain text.
Weekly Content Digest
Founder's analyst
2 tasks

Published posts, impressions and reactions by pillar and format, the best post, engagements that became leads, and what to write more of.

PostsIdeasEngagementsListeningTurtle Notify
  1. 01Count from the tables: Posts Published in the last 7 days by Channel; Impressions, Reactions and Comments summed by Pillar and by Format over the last 30 days; the single post with the most reactions per impression; Engagements with ICP Match Yes and Status New; Ideas Picked with no Posts row; Posts Ready to post with no URL. Output: The counts, as plain text lines.
  2. 02Write the digest: the numbers as short lines, "Best post:" with why, "Leads from content:" the ICP Yes engagements, "Stuck:" the picked ideas and unposted texts. Then "Write more of:" at most three lines. Under 250 words, plain text. Send it with Turtle Notify and confirm. Output: The digest text and a confirmation that it was sent, plain text.
Ideas
One row per idea, by pillar, with where it came from. Picking one drafts it.
7 cols
Idea textPillar selectAngle textSource selectEvidence URL urlStatus selectDrafting agent_output
Posts
Every draft, one row per channel variant. Approving a post formats it and hands you the text; you post it by hand and paste the link back.
16 cols
Post textIdea foreign_keyPillar selectChannel selectFormat selectDraft textStatus selectPublish At dateURL urlImpressions numberReactions numberComments numberLeads numberFinal Text agent_outputRewrite agent_outputHarvest agent_output
Listening
What buyers are asking this week, found by the listener on Reddit, LinkedIn, X and YouTube. The best become ideas.
8 cols
Item textChannel selectQuery textURL urlSummary textWhy It Matters textFound dateStatus select
Engagements
Who reacted to and commented on what you published, with the people who match your ICP flagged as leads.
8 cols
Engagement textPost foreign_keyWho textTitle textLinkedIn urlWhat They Said textICP Match selectStatus select
Newsletter Issues
One row per issue. Adding one assembles the draft from the week's posts into your email tool; you review and send there.
6 cols
Issue textDate dateTheme textAssembly agent_outputStatus selectDraft Link url
Repurpose
Paste a link to a blog post, a doc or a call transcript. The desk turns it into five ideas across your pillars.
5 cols
Source textURL urlKind selectIdeas Found agent_outputStatus select
Voice and Pillars

The founder's voice, the content pillars, the audience and ten posts that worked. Replace the starter text with your own.

2 sources

How the install works

  1. Overview. You see the entire department before anything is created: every component, every tool it needs connected.
  2. Setup questions. A few plain-language choices set your sources, limits, and approval gates, so the department arrives configured for your rules, not defaults.
  3. Conflicts. Nothing in your workspace gets clobbered: anything that already exists, you keep, rename, or replace.
  4. Install. Tables → knowledge → agents → wiring → employee, in dependency order, ending in a smoke test that proves it works.
  5. Done. A working department, not a project plan. Open your new coworker and assign the first task.

Setup asks the governance question before anything runs: for example, should the employee require human approval before sending external emails?

Built on

Connects to

ScrapecreatorsMonid Apify LinkedinSerperMailchimpTurtle Notify

Governance defaults

  • Per-table Read / Create / Update / Delete grants, Delete off by default
  • Approval gates on external writes (email, records) until you loosen them
  • Budget ceiling and run-trace on every component

More like this

All packs