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-W36-Wedkind longformweek W36date 2026-09-09campaign longform-youtubepillar it_devbeat asset videoduration 178.1sground blackscenes 8

Checklist the per-video bar — engine/sim

100.0/100
plain languagevo coverageno dead airuniquenesscaption fitcompletenesscleanliness
quantitative quality · weights learn from your reviews (engine.sim.memory review longform-W36-Wed good|bad)
✓ all static checks pass — one-focal/scene · tier-by-beat · one-track caption · colorway · cast+shape correct · no banned/fabricated. (audio + visual tiers verify on the rendered finals — Phase 2)

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 8 scenes · 178.1s · comp_id + rendered still + tier + the script

#Layer (comp_id · still · tier)BeatTimecodeMotionLogoAudioVO / on-screen / caption
1s1
matches intent
shared field
signature-3d
open0–22.0sspatial-parallaxicon·white-knockout♪ bed_in
Content strategy that moves the needle only works if the needle is pointed at something real. Today is about social listening: pulling what your audience actually says instead of guessing. You will learn how to gather real signal and let it shape what you make. Guessing is cheap and usually wrong. Listening is a little work and almost always right.
on-screen: Stop guessing what they want
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→Stop guessing what they want
2s2
matches intent
NumberedList
template
teach22.0–44.3skinetic-buildicon·wireframe♪ node_lock
Step one. Go where your customers complain, not where they perform. That usually means a forum or a comment section, not a polished feed. We point a scraper at the places our buyer gathers and pull the raw posts and replies. The goal is their words, in their order, before anyone cleaned them up. Complaints are honest. Marketing copy is not.
on-screen: Listen where they complain
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→Listen where they complaincurate items, nodes
3s3
matches intent
ChecklistCard
template
teach44.3–68.1skinetic-buildicon·wireframe♪ node_lock
Step two. Grab the text with an actual tool, not by hand. We use an Apify scraper through an MCP connection, or a Reddit reader, and tell the agent to fetch the top threads on our topic. It returns the posts and the comments as plain text. Now you have a pile of real human sentences instead of a hunch about what people care about.
on-screen: Pull the threads with a tool
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→Pull the threads with a toolcurate items, nodes
4s4
matches intent
NodeGraphCard
template
teach68.1–90.8skinetic-buildicon·wireframe♪ node_lock
Step three. Feed that pile to the model and ask one question: what problem comes up again and again, in their own phrasing. The model is good at spotting the repeat. It hands you back the five complaints that keep returning and the exact words people use. Those words are gold, because they are the words your customer will search and recognize.
on-screen: Let the model find the patterns
expected on screen: black ground · a NodeGraphCard panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id NodeGraphCardvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Let the model find the patternscurate hub, nodes
5s5
matches intent
CodeWindow
template
teach90.8–112.1skinetic-buildicon·wireframe♪ node_lock
Step four. Take the top repeated complaint and make it your next post, answered plainly. Not your pitch, their question, answered straight. We drop the real phrasing into the post so it sounds like we were in the room. You are no longer inventing topics. You are returning the audience's own questions to them with an answer attached.
on-screen: Turn complaints into topics
expected on screen: black ground · a CodeWindow panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id CodeWindowvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Turn complaints into topicscurate codeLines, nodes, windowTitle
6s6
matches intent
AnnotatedDiagram
template
teach112.1–133.4skinetic-buildicon·wireframe♪ node_lock
Step five. Listening does not stop after you post. Pull the replies your post earns and feed those back in. We collect the comments and let the model sort them into questions, agreement, and pushback. The pushback is the best fuel. It tells you the next post before you have to think of it. The loop feeds itself.
on-screen: Close the loop with replies
expected on screen: black ground · a AnnotatedDiagram panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id AnnotatedDiagramvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Close the loop with repliescurate callouts, nodes
7s7
matches intent
StatScoreboard
template
proof133.4–154.70000000000002sreceipts-counticon·wireframe♪ node_lock
The honest proof: before we wrote a single line for the studio, we read what people actually say about agencies that hide the work. That complaint, that you never see how it is made, shaped this whole open build. We did not decide to be transparent because it sounded nice. The audience told us to, and we listened.
on-screen: We listened before we built
expected on screen: black ground · a StatScoreboard panel over a dimmed Signal Field · Faber leads · receipts · receipts-count · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id StatScoreboardvisual receiptsshape boxground blacktreatment wireframemotion receipts-countpower receiptsinstrument spotlight→We listened before we builtcurate pillar, stats, statsLabels
8s8
matches intent
shared field
signature-3d
resolve154.7–178.1scoalescenceicon·white-knockout♪ bed_out
So the takeaway: stop guessing and go read. Scrape where they complain, pull it with a tool, let the model find the repeat, and answer their real words. Then read the replies and do it again. Your next step: see how listening shaped our whole plan at office dot temerarii dot xyz, then go read one thread in your own corner and answer it.
on-screen: Hear the audience, then build
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→Hear the audience, then build

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)

youtubeSocial listening: pull what your audience actually says instead of guessing Guessing what people want is cheap and usually wrong. Listening is a little work and almost always right. This is how to gather real signal and let it shape what you make. The method: - Listen where they complain, not where they perform. That usually means a forum or comment section, not a polished feed. Complaints are honest; marketing copy is not. - Pull the threads with a tool. Use an Apify scraper through an MCP connection, or a Reddit reader, and have the agent fetch the top threads on your topic as plain text. - Let the model find the patterns. Ask one question: what problem comes up again and again, in their own words. It hands back the five complaints that keep returning and the exact phrasing people use. - Turn complaints into topics. Take the top repeated complaint and make it your next post, answered plainly. You are returning the audience's own questions to them with an answer attached. - Close the loop with replies. Pull the comments your post earns and feed those back in. The pushback is the best fuel; it tells you the next post. The honest proof: before we wrote a single line for the studio, we read what people say about agencies that hide the work. That complaint shaped this whole open build. The audience told us to be transparent, and we listened. See how listening shaped our plan at office.temerarii.xyz, then go read one thread in your own corner and answer it. Keywords: social listening, Apify scraper, Reddit research, audience research with AI.

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