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-W44-Satkind longformweek W44date 2026-11-07campaign longform-youtubepillar multimediabeat asset videoduration 132.1sground whitescenes 7

Checklist the per-video bar — engine/sim

98.2/100
plain languagevo coverageno dead airuniquenesscaption fitcompletenesscleanliness
quantitative quality · weights learn from your reviews (engine.sim.memory review longform-W44-Sat good|bad)
⚠ 1 flag(s) — not yet ship-ready: copy_generic · 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 7 scenes · 132.1s · comp_id + rendered still + tier + the script

#Layer (comp_id · still · tier)BeatTimecodeMotionLogoAudioVO / on-screen / caption
1s1
matches intent
shared field
signature-3d
open0–18.8sspatial-parallaxicon·wireframe♪ bed_in
Today closes the week on social media management, but the part people skip: the replies. Building presence in public is not just posting. It is answering. A feed that talks back keeps people. A feed that ignores them dies quiet. Here is how we handle the inbox without living in it.
on-screen: The reply is the real work
expected on screen: white ground · octa hero in the shared Signal Field · Lumen leads · node-graph · spatial-parallax · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual node-graphshape octaground whitetreatment wireframemotion spatial-parallaxpower summoninstrument summon→The reply is the real work
2s2
matches intent
NumberedList
template
teach18.8–38.3skinetic-buildicon·color♪ node_lock
First step. We do not refresh apps hoping to catch a comment. We wire a tool into the agent that pulls recent mentions and replies for our accounts in one place. The model gathers everything that needs an answer into a single list, so nothing falls through the cracks of five separate apps.
on-screen: Pull mentions through a tool
expected on screen: white ground · a NumberedList panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·color logo · caption bottom-left
spec (the prompt): comp_id NumberedListvisual node-graphshape octaground whitetreatment colormotion kinetic-buildpower morphinstrument morph+laser→Pull mentions through a toolcurate items, nodes
3s3
matches intent
ChecklistCard
template
teach38.3–57.8skinetic-buildicon·color♪ node_lock
Second step. Not every reply deserves the same effort. We ask the model to sort the list: the real questions that need a careful answer, the quick thank-yous, and the noise to ignore. Now you spend your time on the few replies that matter instead of scrolling past the ones that do not.
on-screen: Sort what needs a human
expected on screen: white ground · a ChecklistCard panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·color logo · caption bottom-left
spec (the prompt): comp_id ChecklistCardvisual node-graphshape octaground whitetreatment colormotion kinetic-buildpower morphinstrument morph+laser→Sort what needs a humancurate items, nodes
4s4
matches intent
ProcessFlow
template
teach57.8–76.9skinetic-buildicon·color♪ node_lock
Third step. We let the model draft replies in our voice, using the brand rails, but it never sends on its own. You read each draft, fix the tone, and hit send yourself. The model does the typing. You keep the judgment. A real person is always the one who actually replies.
on-screen: Draft answers, never auto-send
expected on screen: white ground · a ProcessFlow panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·color logo · caption bottom-left
spec (the prompt): comp_id ProcessFlowvisual node-graphshape octaground whitetreatment colormotion kinetic-buildpower morphinstrument morph+laser→Draft answers, never auto-sendcurate nodes, steps
5s5
matches intent
NodeGraphCard
template
teach76.9–94.2skinetic-buildicon·color♪ node_lock
Fourth step. When a question gets a strong answer, that answer is content. We drop the good ones back into the source file as next week's posts. A reply you wrote once for one person becomes a post that helps a hundred. The inbox feeds the calendar.
on-screen: Good answers become posts
expected on screen: white ground · a NodeGraphCard panel over a dimmed Signal Field · Lumen leads · node-graph · kinetic-build · icon·color logo · caption bottom-left
spec (the prompt): comp_id NodeGraphCardvisual node-graphshape octaground whitetreatment colormotion kinetic-buildpower morphinstrument morph+laser→Good answers become postscurate hub, nodes
6s6
matches intent
StatScoreboard
template
proof94.2–112.6sreceipts-counticon·color♪ node_lock
The honest proof. We run our own replies this way, drafts from the model, sent by a human, the best ones recycled into posts. You can see which posts started as replies in the calendar at office dot temerarii dot xyz. The loop is real, and it is in public.
on-screen: We answer with the same setup
expected on screen: white ground · a StatScoreboard panel over a dimmed Signal Field · Lumen leads · receipts · receipts-count · icon·color logo · caption bottom-left
spec (the prompt): comp_id StatScoreboardvisual receiptsshape octaground whitetreatment colormotion receipts-countpower receiptsinstrument spotlight→We answer with the same setupcurate pillar, stats, statsLabels
7s7
matches intent
shared field
signature-3d
resolve112.6–132.1scoalescenceicon·wireframe♪ bed_out
The takeaway, and the week's. Presence is a conversation, not a broadcast. Pull your mentions with a tool, sort them, draft with the model, send as yourself, and turn the best answers into posts. Set up your reply list this week, and see the whole loop running at office dot temerarii dot xyz.
on-screen: Answer the room, feed the calendar
expected on screen: white ground · octa hero in the shared Signal Field · Lumen leads · coalescence · coalescence · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual coalescenceshape octaground whitetreatment wireframemotion coalescencepower coalescenceinstrument coalescence→Answer the room, feed the calendar

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)

youtubeHow to prove your production is real with an in-the-machine pipeline Anyone can claim they make their own content. This video shows how to prove it, with an honest, in the machine pipeline you can watch run, so the proof is the working thing and not a promise. The method: Build from one source file in the open. The brief, the brand rules, and the plan all live in files anyone can read. When the inputs are visible, the output is verifiable, because there is nowhere to hide a secret team. Render in code, not a black box. Use Remotion so the video is code a model can read, change, and re render. A timeline in a closed app proves nothing. Code in a repo shows exactly how the piece was made. Use real APIs for the parts you do not film. The voice from ElevenLabs, the stills from an image model, the captions from Whisper. Each one is a file you control and can show, not a mystery step. Check the work with code and show the checks. A banned word scan, a color check, a logo only rule, all running before anything ships, and visible when they run. The proof is not that we say it passed, it is that you can see the gate. Refuse fabricated numbers. Every claim is the method or a real receipt, never an invented result. The honesty is the point, because a made up number is the fastest way to lose the trust the whole thing is built on. The whole calendar you are watching came out of this pipeline, in public, files and all. That is the proof, the working machine you can inspect. See it run at office.temerarii.xyz. Keywords: content production, AI pipeline, Remotion, ElevenLabs, Whisper, transparency, in the machine.

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