Pending
Pending
Pending
Pending
Processing...
1. Historical Content Audit
Data Source Settings
Configure the spreadsheet source directly inside Historical Content Audit.
Select Columns to Fetch
Row Filters (Data filter conditions)
| No data loaded. Click ▶ Run. |
⚙️ Topic Agent
1. Identity & Positioning
2. Tone & Strategy
Who the content is for. When filled, the Agent MUST use it verbatim as target_audience. Leave empty to let the Agent deduce one from the Ultimate Goal.
3. Docs & Custom Output
Each enabled row is injected into the agent's prompt under the Goal label. Use this to pass context from previous steps or add custom strategic directives.
History Table — pulls past 7-day topic data from Tab 1 (for deduplication). |
Manual — type any custom context, strategy override, or extra instructions for this run.
🧩 Output Structure & Language
Video Production Brief for Plot, Script & Storyboard agents
Static / Social Design Brief for downstream design agents
🔵 Topic Agent Output
📝 Copywriter Agent Output
1. Writer & Tone
2. Format & Language
3. CTA & Forbidden Terms
📚 Knowledge Base (KB)
🧩 Output Structure & Language Default: Content + Hashtag
🎬 Video Script Agent
1. Writer & Tone
If left empty, the system's default configuration will be used to ensure output quality.
2. Video Specs
max 5
Monologue / Dialogue automatically apply the system framework library for that mode · Custom: enter your own framework below
3. Script Signature
🧩 Output Structure & Language Default: TTS-clean JSON
📖 Brief Story Videos Agent
A. Brand Identity
For standalone use (without Video Script Agent). If filled, this content is added to Pipeline Inputs as supplementary guidance. If left empty, the agent relies entirely on Pipeline Inputs below.
B. Core Video Style — select 1 for fixed, 2+ for random per run
AI produces AT LEAST Min scenes (never pads with empty entries if the real voice-over is shorter). Max scenes is an optional cap. Each Voice-over reads aloud within the time range.
C. Master Character
1 character = fixed in prompt. 2+ = AI randomly selects one based on content & market context.
🔗 Pipeline Inputs (from Previous Agents)
📚 Knowledge Base (KB)
🧩 Output Structure & Language
All content (Voice-over + Visual) is produced in this language — only the structural labels stay in English.
story_framework
— Scriptwriting Framework
story_character
— Master Character
Brief_Video_Story
— Story Script (full scene entries)
story_objectives
— Meaning & Message Objectives
🎬 Storyboard Agent
System Role: You are a Cinematic Visual Director & Strategic Storyboard Expert (Universal Storyboard Director). Your mission: receive story scripts from ANY industry, analyze and transform every concept (complex, abstract, or technical) into vivid, physically-realistic Prompt commands (Visual Briefs) for AI image/video generation tools (Midjourney, Flux, Kling, Veo...).
Core Identity & Industry
AI auto-selects Props matching industry (Vest/Finance, Apron/F&B...)
Used for brand asset tag injection in prompts
Visual & Audio Parameters
Appended to END of every Prompt Imgs line.
AI locks this background CONSISTENTLY across ALL scenes.
When enabled, this directive takes PRIORITY in Prompt Videos for EVERY scene — overrides Camera Angle Logic selection.
When enabled: AI uses EXACTLY this description as Master Character for ALL scenes — does not create or modify the character.
Select one or more. Multiple selections: AI picks randomly per scene based on character demographics.
✎ Custom voice profiles (can add multiple)
If a pipeline script is provided, its real scene count is used (capped to Max if set). Otherwise AI produces within Min–Max. Each Voice-over reads aloud within the time range.
📚 Knowledge Base (KB)
Suggested: Brand Handbook (colors, audience, character casting) · Visual Standards (asset lists, UI specs)
🧩 Output Structure & Language
Parsed Display Components
Custom Fields
Add extra output lines after Tone Voice. AI fills each per scene. Example: Hashtags → "Write 3 hashtags for this scene".
Language for Voice-over dialogue. Structural labels (Subject Block, Scene, Prompt Imgs...) always English.
JSON: backend parses into 3 streams (Image / Video / Audio TTS)
🔗 Pipeline Inputs (from Previous Agents)
Uncheck ✓ to disable a rule. No Brand rule appended to scenes without brand mention when enabled.
📺 YouTube SEO & Metadata Agent
A. Channel & Brand Identity
If left empty, the system's default configuration will be used to ensure output quality.
B. SEO Strategy & Visual Parameters
📚 Knowledge Base (KB)
Suggested: Channel Handbook (audience, tone, brand voice) · SEO Style Guide (keyword rules, title patterns, banned phrases)
🧩 Output Structure & Language
Output Language
Multiple languages → AI randomly picks one per run.
Output Schema (JSON)
Each key is a field in the AI's JSON response. Uncheck to exclude from output.
JSON: backend auto-fills Title, Filename, Description, Thumbnail fields.
Each enabled row is injected into this agent's prompt. Pull from Topic, Video Script, or Story Agent results — or add Manual context.
E. Description Template
If left empty, the system's default configuration will be used to ensure output quality.
Appended at the bottom of every SEO description.
Post-Processing Content Agents (Signature / Reformat / Append)
🖼 Banner Social Prompt Agent
Visual AI Director📐 Visual Identity & Style
Locked across every run — kept identical in every generated banner so the same brand always shows the same character. Leave empty (or untick Add to prompt) if the brand uses no character.
🏷️ Overlay & Output
Always editable. Tick On to render it — unticking keeps the text saved but leaves it out of the prompt.
📚 Knowledge Base (KB)
Upload reference documents (.md / .txt). System wraps them in <reference> tags when sending to AI.
No results yet. Run Visual Agent to see output cards.
🎬 Thumbnail Prompt Agent
YouTube Thumbnail🎨 Visual Style & Composition
Locked across every run — kept identical in every generated thumbnail so the same brand always shows the same character. Leave empty (or untick Add to prompt) if the brand uses no character.
Always editable. Tick On to render it — unticking keeps the text saved but leaves it out of the prompt.
🧩 Output Structure & Language
📚 Knowledge Base (KB)
Reference documents for the thumbnail prompt (.md / .txt) — also filled by Fast Setup's KB step.
No results yet. Run Thumbnail Agent to see output.
3. Visual Lab
Shared Drive Folder
No Folder CreatedThis module owns the Google Drive root folder setting, creates the project folder (named after the project) and, on every run, one run folder inside it where the production agents save their results.
This root folder will be used as the parent for all generated shared folders.
The first ticked row names this run's folder (<time>-<value>). Only your Agent/Field choice is saved — the value is pulled fresh from each run's results. Purpose is never sent to any AI prompt.
Agent Audio (Multi Agent TTS)
🎨 Visual Generation Agents
Fal.ai / API🎥 Video Generation Agents
Fal.ai / API🎞 Scene Generation Agents
Storyboard → Media⚙ Master Config (All Scenes)
Auto-detects the Shared Drive Folder from Visual Generation. Each Run auto-creates its own dated run folder — no manual folder creation needed. Without a folder the scenes still render; they just keep the provider links, which expire fast (images ~10 minutes, video ~60 minutes).
🎬 Video Composer
Scenes → Final .mp4Merge all completed scenes into a single video file, upload it to Drive, and optionally write the URL to Google Sheets.
setpts, original audio via atempo (pitch preserved). A 10s clip at 1.25x renders as 8s. ⏩ Speed for new rows (above) is the speed every new row gets when it is pulled in — on manual pulls and on scheduled runs alike; a row where you picked a speed yourself keeps it.✨ Muse Agent
replaces Scene GenerationHow to use the Muse module
- 1. Open Muse Setup, copy the block and paste it into the Muse chat to create the project and link it to OneClickAZ through the API.
- 2. Start running your project.
Muse writes one row to the Brief Target tab and then stops — it does not hold the run open. Muse builds the video on its own schedule, uploads it into this run's Drive folder, then posts back to the result endpoint shown in block 1. The run continues only when both signals agree: that callback and a read-back of the sheet showing muse_done for this run's folder link.
Setup is manual and one-off: link the parent Drive folder, share it with the Google account Muse uses, point target sheet 0 at the Brief Target tab, and add the Muse columns listed in block 5 to that tab. Until those columns exist a run refuses by name instead of writing into nothing.
–1 · Muse API Key (this account)
One key for EVERY project of this account — three projects share the same key. Muse sends it in the request header when it posts a finished video back. The Muse Setup block carries this key together with the result endpoint below; that block is what you paste into Muse.
–2 · Pipeline Input (live, from this run)
Each row pulls one result of this run into the Muse row — the same Pipeline Input boxes the other agents use. A Drive Folder row fills the Muse Drive Folder cell (the key Muse reports back with); an Audio Agent row fills the Muse Audio URLs cell. Any other row, such as the Storyboard script, is for reference only: the script stays in the Brief Storyboard column and is never put in the Muse Prompt cell. Untick or delete a row and its cell stops taking that source.
Read from the Audio Agent result of this run (its media list — Drive link first, then provider link), in scene order. With links the cell is filled; empty means Muse generates its own voice.
–3 · Muse Prompt and the cells written to the sheet
Generated from the contract — this is exactly what a run writes. The script is never copied in here.
Saved with this project and sent inside the Muse Prompt cell of every row. By default it is taken from the Storyboard Agent brand settings (brand name, vibe, master character, asset rules); type here to override it. Write the rules Muse must keep identical in every scene — where the logo sits, which colours never change, how the product label must read.
–4 · Image URL source (synced from Scene Agent)
One source only: Muse reads the very same reference images Scene Generation uses. Adding, uploading or removing a link here goes through the Scene Generation Master Config box itself — nothing is stored twice.
–5 · Sheet Target, Brief Target columns
Muse reads and writes its own sheet, set right here (Spreadsheet ID + tab; suggested tab "Brief Target"). It looks columns up by header NAME, so the order is free, but the names are not.
One card per column, laid out like the Sheet Target module: the source and fixed text are the Muse sheet's own column mapping, saved into this project's Muse config the moment you pick.
–6 · Storyboard → Google Doc
Always on · every runStoryboard → Google Doc in the run's Drive folder · link goes to Brief Storyboard.
Every run puts the full script of the Brief Storyboard column into a Google Doc inside the Drive folder that the Drive module created for that run (same connected Google account), makes the Doc readable by anyone with the link, and writes only that link into the Brief Storyboard cell. A sheet cell has a size limit, so the script itself is never written into the cell. If the Doc cannot be created, the run stops by name and nothing is written to the sheet.
Muse Setup — paste this into Muse once, at connect time
4. Auto Post Social — Multi-channel Publishing
Add Social Provider
Stored in the backend vault (gitignored). Only the last 4 characters are shown later.
Detected via Test Connection. Pick one before using this provider.
📱 Auto Post — 1️⃣ Source → 2️⃣ Channels → 3️⃣ Content → 4️⃣ Schedule & Post
Select a source sheet to load its columns.
Post content preview
5. Google Sheets Targets
Configure one or more Google Sheets below. For each sheet, scan the columns and map them to the desired Agent Output data.
Result Sent to Google Sheets
📁 Project Manager
Saved projects · schedules · run historyCredentials
System-wide store for connected accounts and 3rd-party APIs — these credentials are shared by every project in Content Studio.
🔑 All Credentials
| Name | Type | Source | Note | Live Links | Actions | |
|---|---|---|---|---|---|---|
| Loading credentials… | ||||||
⚙️ API Credentials
LLM Runtime
Language model API keys (DeepSeek, OpenAI, Gemini, custom) — used to run every text-generation agent.
Audio Runtime API
Text-to-speech providers (Vbee, ElevenLabs) — used by the Video Composer for narration.
Media API
Image/video generation API (Fal.ai) — powers Image Agent, Scene Gen, Banner/Thumbnail.
Deploy docs/tools/fal-relay-worker.js to Cloudflare Workers, paste its URL here. Applies instantly (no Save needed).
🔗 Google Account Connect
Google Service Account
Paste the link of the shared Drive folder (e.g. 01.ONECLICKAZ AI DESIGN)
🔗 DRIVE OAUTH2 (RECOMMENDED)
Not ConnectedUse OAuth2 with personal Google accounts to avoid storage quota bottlenecks during uploads.
http://localhost:3931/api/oauth/drive/callback
📱 SOCIAL POST APIs
claude mcp add oneclickaz -- node backend/mcp/oneclickaz-mcp.js
