# How the Content Workflow Works This toolkit helps you go from “I want to make better short-form content” to organized research, recorded footage, and finished videos. You do not have to use every skill every time. Think of it as a small set of helpers you can call when you need them. The first time you use one, it explains what it does and asks how you like to work. It saves those preferences in your project so you do not have to start from scratch every time. ## First, install the skills - Unzip `content-workflow-toolkit.zip`. - Copy the five skill folders into your Codex skills folder, usually `~/.codex/skills/`. - Start a new Codex task inside the project where you want to keep your content work. - Call a skill by typing its name with a `$`, like `$creator-research-scraper`. If you only want one piece of the workflow, you can install the individual `.skill` file instead of the whole bundle. ## Here is the quick version - Research the creators or accounts you want to learn from. - Put the useful posts into a Google Sheet. - Choose an idea and prepare your hooks and shared body. - Load the approved copy into the teleprompter. - Record all of the hooks and the body. - Cut the raw recording into clean hook clips and one clean body clip. - Edit and approve the body once. - Edit and approve each hook separately. - Attach every finished hook to the same finished body. - Optionally create a useful free resource and connect it to a Blotato comment-and-DM funnel. - Review the final videos, then publish them separately when you are ready. That “body once, hooks separately” order is important. If you have five hooks, you should not edit the same body five different times. You finish the body once, reuse it, and only customize the opening hook for each version. ## Step 1: research content that is already working Start with `$creator-research-scraper`. Tell it which Instagram or TikTok accounts you want to research and how far back you want to look. It collects the posts and turns them into one consistent research file instead of leaving you with a pile of random links. The skill will ask for your watchlist, platforms, date range, post limit, and budget preferences. If the scrape may cost money, it shows you the planned run before starting. You can say: > Use $creator-research-scraper. Explain how it works, then help me research these accounts from the last 30 days. ## Step 2: organize the research in Google Sheets Next, use `$research-google-sheet`. This takes the research file and puts it into a Google Sheet with the creator, platform, post date, views, engagement, caption, source link, thumbnail, and blank spaces for your hook, body, and adaptation notes. You choose the spreadsheet and worksheet. You can add new rows, replace one worksheet, or update existing posts without duplicating them. The skill previews the number of rows and columns before it writes anything. If you plan to offer free resources later, ask it to include the optional resource-funnel columns. Those columns let you track the guide title, URL, comment keyword, Blotato setup, follow gate, DM copy, and verification status. You can say: > Use $research-google-sheet to organize this research. Show me the columns first, and include the optional free-resource and Blotato tracking fields. ## Step 3: choose an idea and prepare the script Look through the Sheet and choose the idea you want to make your own. Write the script in your own voice and make sure any factual claims are supported. The video setup works especially well when you record several different hooks that all lead into one shared body. For example, you might write five openings for the same lesson, then record the lesson once. This bundle expects the hooks and shared body to be approved before they go into the teleprompter. It does not automatically copy another creator's script or branding. ## Step 4: put the script into the teleprompter Use `$teleprompter-builder`. Give it your approved hooks and body. It asks how large you want the words, how fast you read, whether the screen should be mirrored, and which controls you want. Then it builds a local teleprompter that opens in your browser. You can say: > Use $teleprompter-builder to turn these approved hooks and body into a teleprompter. Ask how I want it to look before you build it. ## Step 5: record the hooks and body Record all of the hooks and the shared body. It is fine to do multiple takes. Leave a small pause between takes so the cutter can recognize the sections more easily. Keep the original recording. The toolkit is designed to make new working files rather than replace your raw footage. ## Step 6: cut the raw recording into clean pieces Use `$hook-body-video-cutter`. It transcribes the recording, finds the likely takes, and proposes a cut plan. You review the choices before it renders anything. Once approved, it exports one clean body clip and a separate clean clip for each hook. You can say: > Use $hook-body-video-cutter on this recording. Show me the proposed body and hook takes before you render the clean clips. ## Step 7: finish the shared body first Now use `$hook-body-reel-stitcher`. The name says “stitcher,” but it also handles the full edit. Before it starts, it asks about your hook style, captions, fonts, colors, animation energy, browser captures, GIFs or reactions, sound effects, music, video size, and closing CTA. The body gets edited first. This is where most of the work happens: - word-by-word captions; - animations and visual explanations; - current website or software screenshots; - short browser recordings when the actual interaction matters; - GIFs or reaction clips when they help the joke or emotion; - logos, diagrams, and proof visuals; - sound effects that match visible actions; - music and audio mixing, if you want them. Codex shows you the finished body for review. The body has to be approved before the final versions are assembled. ## Step 8: finish each hook Once the body is approved, the skill edits each hook separately. Each hook gets its own opening text treatment, word-by-word captions, and any hook-specific animation or sound. You approve every hook on its own. Approving one hook does not automatically approve the others. ## Step 9: stitch the finished pieces together After the body and hooks are approved, the skill creates the final versions: - Hook 1 plus the finished body - Hook 2 plus the finished body - Hook 3 plus the finished body - and so on It checks the hook-to-body join, audio, captions, dimensions, frame rate, missing assets, frozen frames, and the beginning and end of every video. You can say: > Use $hook-body-reel-stitcher with these clean clips. Ask for my editing preferences, finish and show me the shared body first, then finish the hooks, and only assemble the final videos after I approve the components. ## Step 10: optionally add a free resource and Blotato funnel You can skip this entire step. Use it when the post says something like, “Comment GUIDE and I’ll send it to you.” First, create something genuinely useful for the viewer—a checklist, template, prompt pack, worked example, decision guide, or source list. Put it on your own website and test the live page on both phone and desktop. Make sure the page is complete and every link, download, form, or copy button works before you promise it in a post. Then choose a short, unique keyword and add the CTA to your content. In Blotato, connect the exact Instagram account you want to use and create two matching paths: - one for someone who comments the keyword; - one for someone who sends the keyword by DM. Both paths should point to the same resource and use the same promise. Both should also include a separate follow gate asking the person to follow the selected account and tap an “I’m following” button. That button is a self-confirmation; it does not actually verify the follow. After the follow step, send a friendly DM that names the resource, says what it will help with, and includes a button to open the verified website page. Do not repeat the follow request in the delivery message. Build the automations as drafts first. Check the account, keyword, message copy, button, URL, follow gate, and active status before turning them on. Then test both the comment path and the DM path from a different Instagram account. The full checklist and fill-in-the-blank message examples are in [OPTIONAL_RESOURCE_FUNNEL.md](OPTIONAL_RESOURCE_FUNNEL.md). The toolkit explains and tracks this optional branch, but website deployment and Blotato changes still require separate approval. ## Step 11: do one final review Watch every finished video all the way through. Check the captions, audio, hook-to-body transition, screenshots, GIFs, animations, sound effects, CTA, and ending. If you built a resource funnel, also open the live resource link and test both Blotato trigger paths. Do not publish the keyword CTA until the resource and delivery flow actually work. Uploading, scheduling, and publishing are separate actions. The toolkit will not do those automatically just because the video is finished. ## What you need - Python 3 for the included helper scripts. - An Apify token if you use the included scraping route. - Connected Google Sheets access if you want the Sheet written for you. - A browser for the teleprompter and website captures. - FFmpeg and a transcription tool for cutting video. - Node.js 22+ and HyperFrames for the complete video edit. - Your own website project and hosting access for an optional free resource. - Your own connected Instagram account and Blotato access for the optional comment-and-DM funnel. The toolkit does not include anyone else's creator list, account IDs, credentials, website, branding, voice, or private files. Each person supplies their own information and decides which external actions to approve.