builderon-demandauto-generated · unproven

How do I take a downloaded VEED video, study reference videos for style, debate exactly where and what B-roll to add, generate those B-rolls with the Runway ML API, and assemble + render the final video in Remotion, returning a link to the output

the agent that answers this

veed B-roll enhancer (Runway + Remotion). Take a downloaded VEED video, study reference videos for style, debate exactly where and what B-roll to add, generate those B-rolls with the Runway ML API, and assemble + render the final video in Remotion, returning a link to the output.

1 runLast run 2d ago1 on a scheduleUpdated Jun 13v4

Cost: Free; on your own plan
Runs: on-demand; on a schedule
Built from: 12 steps; 4 verified skills
Runs in: Claude or Codex; as you

what you get

A rendered, B-roll-enhanced version of your VEED video plus a shareable link to the output file, with the B-roll shot list approved before any rendering happens.

example result

illustrative

A finished on-demand run delivers: A rendered, B-roll-enhanced version of your VEED video plus a shareable link to the output file, with the B-roll shot list approved before any rendering happens.

Example output, to show the shape of the deliverable. It has not run on your data. Build the agent to get this on yours.

the steps

12 steps · 4 from verified skills

Step 1tool
Preflight: ensure the toolchain is installed/idempotent — Remotion (npx), ffmpeg/ffprobe, whisper.cpp + a ggml model for transcription, yt-dlp for the reference, and the Runway API key (RUNWAYML_API_SECRET).
Your model fills this step
- No integration? brew install ffmpeg
- No integration? uvx yt-dlp
- No integration? npm i in the remotion project
Step 2decision
Confirm the exact filename of the downloaded VEED input video and that the file exists before doing anything else. Also confirm the creator's exact on-screen NAME SPELLING (e.g. 'Sanna') — use it verbatim in every graphic.
Decision step
Step 3tool
Sample the SOURCE video's dominant colors and LOCK the design system to MATCH it — never a generic dark theme (this creator = light/white + red + green). Lock the 5 recurring elements: (a) hero compass motif; (b) matched grade; (c) type system (geometric sans for kinetic text + serif for 'ancient wisdom' beats, always animated in, never static lower-thirds); (d) audio plan; (e) SFX kit. Use the confirmed name spelling on all cards.
Your model fills this step
- No integration? ffmpeg palettegen on a few frames
Step 4tool
Open the inspiration reference video(s) and extract the editing language: B-roll style, cut cadence, overlay-vs-hard-cut, motion/energy, color treatment.
Browser automation with persistent page state. Use when users ask to navigate websites, fill forms, take screenshots, extract web data, test web apps, or…
Dev Browser A CLI for controlling browsers with sandboxed JavaScript scripts. Installation npm install -g dev-browser dev-browser install Usage Run dev-browser --help to learn more.
Full skill: dev-browser
- No integration? yt-dlp the reference, build a frame contact sheet, and study it
Step 5
Transcribe the VEED input and build a second-by-second timeline of spoken content, scene changes and natural pause points — INCLUDING the exact timestamps where the speaker says numbered points ('one, two, three, four') so audio dings can land on them.
Automate audio/video transcription, meeting notes, subtitle generation, and content processing
Transcription Automation Comprehensive skill for automating audio/video transcription and content processing. Core Workflows 1. Transcription Pipeline TRANSCRIPTION FLOW: ┌─────────────────┐ │ Audio/Video │ │ Input │ └────────┬────────┘ ▼…
Full skill: Transcription Automation
Step 6
Debate WHERE and WHAT to place into a beat-by-beat map. CRITICAL: every B-roll must match the ACTUAL narrative moment, never a literal keyword match (a flag/government building on a market/story beat is wrong). Retention rule: no static talking-head >6-8s without a cut/overlay/push-in/graphic. Graphics use the matched palette + recurring compass motif.
Your model fills this step
Step 7decision
Present the placement map + locked design system (palette, motif, type, name spelling, audio/SFX plan) for approval or edits before generating anything.
Decision step
Step 8tool
For PHOTOGRAPHIC shots, generate B-roll via Runway (gen4_turbo image-to-video, seed via gen4_image). For TEXT/DIAGRAM shots build them NATIVELY in Remotion in the matched palette with kinetic reveals — never send legible text/diagrams to Runway.
Produce programmable videos with Remotion using scene planning, asset orchestration, and validation gates for automated, brand-consistent video content.
Remotion Video Production A skill for producing programmable videos using Remotion. Generate automated videos from text instructions and produce brand-consistent videos at scale. When to use this skill Automated video generation: Create…
Full skill: remotion-video-production
- No integration? if credits are low, reduce clip count/duration and lean on native graphics
Step 9
Assemble in Remotion (or composite onto a prior render if the clean base is gone): color-grade base to the locked palette, composite each B-roll at its approved timestamp, animate graphics in, thread the compass motif, match the reference pacing.
Produce programmable videos with Remotion using scene planning, asset orchestration, and validation gates for automated, brand-consistent video content.
↳ same skill as step 8 (remotion-video-production)
Full skill: remotion-video-production
Step 10tool
Audio layer. DEFAULT: NO continuous background-music bed unless the creator explicitly asks — a constant bed reads as distracting. Instead: (a) put a SHORT, pleasant ding/click at the start of EACH spoken numbered point (4 points = 4 dings), timed to the number, subtle, non-disruptive; (b) put an appropriate SFX on EVERY B-roll insert on entry (soft whoosh default; fitting ambient where obvious — wind on aerials, keys jingle on a keys shot, sub-bass on number reveals, shimmer on the compass, decode tick on the MLS line); (c) clean the voice (de-noise + gentle compression + warm EQ). If music IS requested, duck it 6-8dB under VO and dip to silence before big reveals.
Your model fills this step
- No integration? generate SFX via Runway eleven_text_to_sound_v2
- No integration? build the SFX/ding filtergraph from a file via ffmpeg -filter_complex_script (avoids shell-quoting breakage; asplit or re-open files for reused inputs)
Step 11decision
Quality gate: critique as domain expert, skeptic, end user. Explicitly verify: graphics MATCH the source palette and read premium (not dull/dark); name spelling correct; NO distracting music bed; a ding lands on each spoken numbered point; every B-roll has an SFX; no literal-but-wrong B-roll. Fix problems in place.
Decision step
Step 12decision
Apply the creator's precise edit notes (exact cut in/out points, e.g. 'remove 2s at 7:00-7:01') by snipping the assembled A/V together so audio + video stay in sync, then render to a final file (default 1080p) and present the output + a shareable link for sign-off.
Decision step

runs hands-free with

scheduled runs
fires this workflow automatically on a schedule so it delivers on its own instead of being re-run by hand
Claude for Chrome
pulls live data from sites that have no API (an MLS, Zillow, your CRM web UI) and clicks through web tasks for you, so the data-gathering steps run themselves instead of leaving placeholders to fill

Connect these and the agent gathers its own data and delivers on a schedule, instead of leaving you blanks to fill.

Build and run this on your own Claude or Codex, free

Install Implexa, then say build the veed B-roll enhancer (Runway + Remotion) agent and approve the schedule. It runs as you, on your real data, on the subscription you already pay for, and gets sharper each run. Your agent's memory is yours and travels with you across Claude, Codex, and whatever comes next. About 5 minutes to your first real run.

one command. signs you up + installs the plugin in claude code. free tier forever.

common questions

How do I take a downloaded VEED video, study reference videos for style, debate exactly where and what B-roll to add, generate those B-rolls with the Runway ML API, and assemble + render the final video in Remotion, returning a link to the output?

The veed B-roll enhancer (Runway + Remotion) agent. A rendered, B-roll-enhanced version of your VEED video plus a shareable link to the output file, with the B-roll shot list approved before any rendering happens.

Is the veed B-roll enhancer (Runway + Remotion) agent free?

Yes. It runs on the Claude or Codex subscription you already pay for, so there is no extra AI bill and no per-run charge. You can build and run unlimited agents on the free plan.

How often does the veed B-roll enhancer (Runway + Remotion) agent run?

It is built to run on-demand, on a schedule you set when you build it. You can change the cadence or pause it any time, and it runs unattended once it is on.

What does the veed B-roll enhancer (Runway + Remotion) agent need to run?

Install Implexa into your Claude or Codex, then connect scheduled runs and Claude for Chrome so it can gather its own data and deliver hands-free. Implexa never touches your accounts or credentials.

Does the veed B-roll enhancer (Runway + Remotion) agent use my data? Is it private?

It runs as you, on your own machine, on your real data. The model runs inside your own Claude or Codex, so Implexa never sees your data, accounts, or credentials. Your agent's memory is yours and travels with you across Claude, Codex, and whatever comes next.

How do I build the veed B-roll enhancer (Runway + Remotion) agent?

Install Implexa into your Claude or Codex, then say "build the veed B-roll enhancer (Runway + Remotion) agent" and approve the schedule. Implexa assembles the 12 steps (4 from verified skills) and it runs on its own. About 5 minutes to your first real run.

Can I change what the veed B-roll enhancer (Runway + Remotion) agent does?

Yes. Tell it what to change in plain language and it revises its steps; the next scheduled run uses the change, with no re-scheduling. Every change is versioned, and a run can even propose its own improvements.

changelog

v4Jun 13manual
rebound step 5 (skills.sh/blitzreels-video-editing -> skills.sh/transcription-automation), step 6 (decision -> gap), step 8 (gap -> skills.sh/remotion-video-production), and more
v3Jun 13manual
rebound step 6 (gap -> decision), step 8 (skills.sh/remotion-video-production -> gap), step 10 (gap -> skills.sh/videoagent-audio-studio)
v2Jun 12manual
added 2 steps; rebound step 3 (skills.sh/dev-browser -> gap), step 4 (skills.sh/blitzreels-video-editing -> skills.sh/dev-browser), step 5 (gap -> skills.sh/blitzreels-video-editing), and more
v1Jun 12generated
auto-generated from "Take a downloaded VEED video, study reference videos for style, debate exactly w"

Agents are alive: every change is a version, and a run can propose improvements that get reviewed and applied.