Back Office · office.temerarii.xyz
One asset, all the way in — composition, the wireframe + storyboard, the output format stack, and the template, all read from the SAME content-index record. The expected output matches what /media surfaces for this post.
post longform-W45-Sunkind longformweek W45date 2026-11-08campaign longform-youtubepillar it_devbeat asset videoduration 229.0sground blackscenes 10

Checklist the per-video bar — engine/sim

97.2/100
plain languagevo coverageno dead airuniquenesscaption fitcompletenesscleanliness
quantitative quality · weights learn from your reviews (engine.sim.memory review longform-W45-Sun good|bad)
⚠ 2 flag(s) — not yet ship-ready: custom_element_dup · see docs/strategy/VIDEO-CHECKLIST.md

Composition comp · template family · expected output

composition LongFormChaptersfamily / template LongFormChapters
9:16 Reelpending1:1 Squarepending16:9 Widepending9:16 4Kpending1:1 4Kpending16:9 4KpendingGIF (SMS)pending
render pending — silent master not yet on disk
expected output: 0/7 rendered — same matrix the /media preview surfaces for this asset.

Composition layer × scene 10 scenes · 229.0s · comp_id + rendered still + tier + the script

#Layer (comp_id · still · tier)BeatTimecodeMotionLogoAudioVO / on-screen / caption
1s1
matches intent
shared field
signature-3d
open0–24.8sspatial-parallaxicon·white-knockout♪ bed_in
This week the calendar says it plain. We are showing you the social engine, the thing that turns strangers into people who buy from you. Most agencies treat social like a mood. Post, hope, repeat. We treat it like a machine with named parts. Today we walk the whole machine end to end, social strategy first, and we hand you every move so you can run it yourself.
on-screen: The whole social engine, in public
expected on screen: black ground · box hero in the shared Signal Field · Faber leads · node-graph · spatial-parallax · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual node-graphshape boxground blacktreatment white-knockoutmotion spatial-parallaxpower summoninstrument summon→The whole social engine, in public
2s2
first render · fix pending
NumberedList
templatecustom_element_dup
teach24.8–50.0skinetic-buildicon·wireframe♪ node_lock
The first move is boring and it matters most. Put your whole plan in one file. We use one calendar file, a plain text source, that lists every week, every pillar, and every post. The machine reads that file. You do not chase a dozen documents. When the brain is The Big T-M running in the terminal, it opens that one file and knows the entire quarter at once.
on-screen: Start with one source file
expected on screen: black ground · a NumberedList panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id NumberedListvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Start with one source filecurate items, nodes
3s3
first render · fix pending
ChecklistCard
templatecustom_element_dup
teach50.0–74.5skinetic-buildicon·wireframe♪ node_lock
Next, choose where you show up. Do not be everywhere. Open Claude Code, list your buyers, and ask the model which two or three platforms those buyers actually sit on. Then you wire each account once through a posting tool. We use Blotato through an MCP server, so one agent posts to every account from the same call. Set it up once. Stop logging into seven apps.
on-screen: Pick the platforms that fit
expected on screen: black ground · a ChecklistCard panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id ChecklistCardvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Pick the platforms that fitcurate items, nodes
4s4
matches intent
DiffCard
template
teach74.5–97.5skinetic-buildicon·wireframe♪ node_lock
Now you write. The model already has the source file, so it knows this week is about the social engine. You tell it the day and the angle. It drafts the posts in your voice, not a generic voice. The trick is to feed it your real past posts as examples so it copies how you actually talk, dry and plain, never loud.
on-screen: Write the week from the plan
expected on screen: black ground · a DiffCard panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id DiffCardvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Write the week from the plancurate lines, nodes
5s5
matches intent
TerminalRun
template
teach97.5–120.2skinetic-buildicon·wireframe♪ node_lock
One idea is enough for a week. Take a single point, like start with one source file, and have the model cut it into a short video script, a long caption, and a thread. Same brain, different shapes. You wire a transcript tool so it pulls the words straight from your video, then rewrites each version for the platform it lands on.
on-screen: Turn one idea into many posts
expected on screen: black ground · a TerminalRun panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id TerminalRunvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Turn one idea into many postscurate nodes
6s6
matches intent
CheatSheet
template
teach120.2–145.7skinetic-buildicon·wireframe♪ node_lock
Before you post more, listen. Point a search tool at your own name and your competitors. We use a Reddit and X reader through MCP to pull what people already complain about. The model reads the threads and tells you the exact words your buyers use. You write back in those words. That is social listening, and it is just reading the room with a machine that never gets tired.
on-screen: Listen before you talk
expected on screen: black ground · a CheatSheet panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id CheatSheetvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Listen before you talkcurate nodes, points, uses
7s7
matches intent
StepFlow
template
teach145.7–166.6skinetic-buildicon·wireframe♪ node_lock
Running accounts by hand eats your week. Instead, have the model take the drafted posts and push them into the scheduler with the right dates straight from the source file. One command fills the calendar. You are not setting alarms. The machine knows Tuesday is white ground and a content post, because the file told it so.
on-screen: Schedule it once, walk away
expected on screen: black ground · a StepFlow panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id StepFlowvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Schedule it once, walk awaycurate nodes, steps
8s8
matches intent
LegacyCard
template
teach166.6–188.6skinetic-buildicon·wireframe♪ node_lock
The engine is not done when you post. Wire the comment feed into the agent. It reads new replies and DMs, flags the ones that sound like a buyer, and drafts an answer you approve in one line. You stay human at the door. The model just makes sure nobody waits three days for a reply that decides the sale.
on-screen: Read the replies, answer fast
expected on screen: black ground · a LegacyCard panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id LegacyCardvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Read the replies, answer fastcurate nodes
9s9
matches intent
ComparisonTable
template
proof188.6–211.29999999999998sreceipts-counticon·wireframe♪ node_lock
Here is the honest proof. We do not have a separate team doing our social by hand. The same engine we just described runs The Big T-M in public. Every post you see from us came out of one source file and one agent in a terminal. No secret room. You are watching the method work on the people who built it.
on-screen: We run our own feed this way
expected on screen: black ground · a ComparisonTable panel over a dimmed Signal Field · Faber leads · receipts · receipts-count · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id ComparisonTablevisual receiptsshape boxground blacktreatment wireframemotion receipts-countpower receiptsinstrument spotlight→We run our own feed this waycurate colA, colB, rows, statsLabels
10s10
matches intent
shared field
signature-3d
resolve211.3–229.0scoalescenceicon·white-knockout♪ bed_out
That is the social engine, start to finish, given away. One source file, one brain, real platforms, listening, scheduling, replies. Nothing here is hidden behind a meeting. Go watch it run at office dot temerarii dot xyz, then build your own. The whole point is that you can.
on-screen: See it live at the office
expected on screen: black ground · box hero in the shared Signal Field · Faber leads · coalescence · coalescence · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual coalescenceshape boxground blacktreatment white-knockoutmotion coalescencepower coalescenceinstrument coalescence→See it live at the office

Format stack 1 aspects · same scenes[], re-cropped

16:9
1920×1080
X/Twitter · YouTube · LinkedIn video

Channels 2 destinations

YouTubeBlog

Social captions supplemental published copy · per channel (comp_id level)

youtubeThe Whole Social Engine, Start to Finish, Built on One File and an AI Agent Most agencies treat social like a mood: post, hope, repeat. We treat it like a machine with named parts. In this video we walk the whole machine end to end, social strategy first, and hand you every move so you can run it yourself. What we cover: - One source file. Put your whole plan in one file that lists every week, every pillar, and every post. The agent reads that file and knows the entire quarter at once. - Pick the platforms that fit. Do not be everywhere. Ask the model which two or three platforms your buyers actually sit on, then wire each account once through a posting tool like Blotato over an MCP server. - Write the week from the plan. The model already has the file, so it knows this week's theme. Feed it your real past posts so it copies how you actually talk, dry and plain, never loud. - One idea, many posts. Take a single point and have the model cut it into a short video script, a long caption, and a thread. A transcript tool pulls the words straight from your video. - Listen before you talk. Point a Reddit and X reader through MCP at your name and your competitors, and let the model tell you the exact words your buyers use. - Schedule it once. Have the model push the drafted posts into the scheduler with the right dates from the source file. One command fills the calendar. - Read the replies, answer fast. Wire the comment feed into the agent. It flags the buyer-sounding replies and drafts an answer you approve in one line. The honest proof: we do not have a separate team doing our social by hand. Every post you see from us came out of one source file and one agent in a terminal. Go watch it run at office.temerarii.xyz. The takeaway: one source file, one brain, real platforms, listening, scheduling, replies. Nothing here is hidden behind a meeting. Keywords: social media strategy, AI agent, MCP, Blotato, content workflow.

Cross-links every lens is a view on this one record