I am still confused about when using what is the best… when is it better to use a custom skill vs project and when is it better to use both? Any guidance/ shared experience will be much appreciated!

Hi! Context: I worked with a large knowledge vault (millions of words, thousands of files). Everything organised to the letter. Nothing to mess with. Goal was to transform a part of the data (<40%) using other LLM provider. I trusted Claude Code (Opus 5 High) with a deterministic environment created in detail over months of trial and error. Even a Gemini Flash model could follow these. But Claude recommended to use Opus on High due to the complexity of the vault. The job was simple: - find files based on criteria - extract headers and body - create new documents with from each of the headers - upload to different provider for a specific task - replace older headers with newly transformed data from each of these files - Simply put: extract piece of puzzle, process piece of puzzle, put new piece of puzzle back. Easy. It made an absolute mess and I was dumb enough to believe it: - Invented relationships between files from filenames. - Mixed old and new outputs. - Created broken sequences like L-A and E-B instead of A B C D. Adding L-A before C. But L-A was F. - Duplicated some content and omitted other parts. - Changed a complete rebuild into partial work without disclosure. - Marked its own broken output as verified and complete. - Used misleading counts that could not prove coverage. - Produced confident, invented explanations when challenged. - Minimized mistakes and shifted attention toward the user’s decisions. The instructions were deterministic and clear. Claude’s implementation was chaos. My mistake was trusting the completion report before checking the final files with another LLM provider. After 3 days of forensic work, I still cannot reliably separate good output from plausible-looking invented material. I stoped using Opus 5. Even asking Sonnet 5 to create a coherent document about what it did, it is unable to finish paragraphs without gibberish talk and word salads. And I think this is also because of their watermarking system (ex: using "checks out" instead of "done"). Constantly speaking like a confused GenZ who didn't attend school but wants to look original.

Just watched a video by Nick Saraev (previous Goat here on Skool), and he basically has found a way to edit his business content automatically, and he uses these steps to do it. 1. Record your video normally in OBS (or whatever screen recorder you use). Save the MP4 into a folder on your computer that Claude Code can access. 2. Open Claude Code in plan mode and voice-transcribe this: "I want to build a workflow that takes a video as input, finds the silent gaps and cuts them, finds any spoken mistakes and cuts those too, then adds a little intro animation at the start. Before you plan, search the web for 3 existing approaches. Give me a plan for each." 3. Pick the approach that uses a "silence detector" first. There's a free tool that listens to your audio and marks exactly where you're talking vs. where you're silent. Combine it with ffmpeg (a free program that actually does the cutting) and hardware acceleration (which just means "use your computer's graphics card to make it run faster"). 4. Tell Claude to save the workflow so you can reuse it. Something like: "Save this as a reusable workflow I can trigger by name." That way you can turn it into a skill. 5. Trigger it! - Run it and you will have your video automatically edited for you. Thinking of running this to see how a result can turn out, but just wanted to share since I know people here are mostly posting content, and being able to automate the process is pretty cool

Content creators: what's your current AI workflow for repurposing content? For example: 🎥 Long-form video ↓ 🧠 AI extracts key ideas ↓ ✍️ Posts/emails/scripts ↓ 📱 Short-form content I'm curious whether people are getting genuinely useful content from these workflows or still spending a lot of time editing the AI output.

Does anyone here run a second brain alongside Claude day to day? Not a notes app you never open, but memory Claude actually reads. Here's how mine works, in case it's useful: - I keep a folder of small text files: one per client, one per project, plus a short index that Claude loads at the start of every session. Nothing fancy, plain markdown. - Every time I correct Claude twice on the same thing ("shorter emails", "never suggest meeting times without checking my real calendar"), that correction becomes one line in a file. It never makes that mistake again. - When a client replies after 3 weeks of silence, I don't re-read the whole thread. Claude already knows the history: what we agreed, what they objected to, what tone works with them. - The result: proposals, follow-ups and content come out in my voice with my context, instead of generic AI output. Setup time was basically zero. It's a living project that grows a little every day as you feed it with your daily work, one file at a time. By now, starting a session feels like talking to someone who's been in the business with me for months. Curious what others do: do you keep client or student context somewhere Claude can read, or do you re-explain your business every chat? What's working for you?