Recent content on Code for Creatives Blog
Two Screen Studio recordings from a C4C cohort session wouldn’t open. The folders were right there on the external drive - “terminal intro pt1” and “terminal intro pt2” - stuffed with 464 and 377 files respectively. Every fragment of the recording was accounted for: video segments, audio, cursor data, keystroke logs, webcam footage. But Screen Studio wouldn’t…
Alex Dobrenko shared this message with the community earlier today: I’m excited to share that we’ve acquired Anthropic. The Anthropic team brings strong engineering instincts, deep model understanding, and a proven ability to build AI that actually works. As we’ve grown Code for Creatives from a small cohort of creative professionals into whatever this is now, it became clear…
Nobody sits down and says “today I will build a skill.” That’s not how it happens. What happens is: someone finds a useful tool, wants to share it with their community, asks Claude to help, and somewhere in the middle of doing that thing, notices it has a shape. A repeatable shape. And then they capture the shape. That’s the whole story. But the details are worth unpacking.…
You install this once. After that, Claude uses it on its own. You don't trigger it, you don't remember to run it, you don't do anything. Everything below happened automatically. Claude Code can write HTML but it can’t see what the page looks like. It doesn’t have a browser. So when it builds you a webpage and says “done,” it’s guessing based on the code alone.…
Plausible shows visitors and page views. Google Search Console shows what people searched to find the site. YouTube analytics live in YouTube. Checking all three means three tabs, three logins, and enough friction that it just doesn’t happen. Claude Code can connect to all three and combine them into a single HTML dashboard. One command refreshes everything. Connecting to Google Search…
Claude Code supports custom agents defined as markdown files in .claude/agents/ . Each agent has a name, a description, a model, and a set of instructions. Any Claude Code session can spawn them. Once an agent exists, it’s available everywhere. The screenshot agent uses that system to capture web pages and local HTML files on demand. Playwright handles the browser automation. Chromium runs…
A full SEO and GEO overhaul on a Hugo blog deployed to Cloudflare Pages covers technical SEO (sitemap, meta tags, structured data, broken links), internal cross-links between related posts, title and description optimization, and GEO bookends (quick answer blocks, definition sections, FAQ sections) that let AI search engines extract and cite the content. Every post gets touched, but the existing…
Claude Code can create presentation slides in two formats: PowerPoint (.pptx) files generated via the python-pptx library, and HTML files with arrow-key navigation and contenteditable text. PPTX works for sending to others who need an editable file. HTML works for presenting from a browser and iterating quickly. Merging new slides into an existing PPTX deck takes multiple attempts due to OOXML…
Claude Code hooks are scripts that run automatically on session events. Four events are available: PreToolUse (before a tool runs), PostToolUse (after a tool finishes), Notification (on status messages), and Stop (on session end). Hooks are registered in ~/.claude/settings.json and execute asynchronously. A practical use case is session awareness: a PostToolUse hook calls Claude Haiku every 20…
Claude Code resets effort level to medium every session because a built-in recommendation overrides saved preferences. The fix is to use the --effort max CLI flag at launch. For a permanent solution, users can create a shell alias like alias c='claude --effort max' in their ~/.zshrc file. The settings.json and /config options save the preference but it gets ignored on startup. What is the Claude…
Git is a version control system that tracks file changes over time. Unlike normal file saving, git takes permanent snapshots called commits that can always be revisited. Branches create parallel versions of a project that can be worked on independently and merged later. Git works on any text file, making it useful for essays, notes, and project plans in addition to code. The mental model is…
Claude Code can batch rename video files based on their visual content by using ffmpeg to extract still frames and analyzing them. For each video, Claude extracts three frames (beginning, middle, end), identifies what’s happening in each, and generates descriptive filenames like 064_graduate-devin-explaining-heat-pump-visitor.MP4 . A dry run shows proposed names before any files change.…
Claude Code can build a full-stack student progress dashboard in one session: a personalized checklist for students, an admin view for the instructor, and a Supabase database connecting them. The build used Claude Code’s plan mode to produce a 10-section plan, then generated seven files covering the database schema, API layer, authentication, and frontend views. Three bugs emerged after…
Claude Code can build editable HTML presentation slides and deploy them to a live URL via Cloudflare Pages in a single session. The slides use contenteditable for in-browser text editing and arrow keys for navigation. The build process involves 8-12 rounds of visual feedback, each taking about 2 minutes. Deployment uses Wrangler CLI with one command: wrangler pages deploy folder --project-name…
Building HTML emails with Claude Code follows a three-phase workflow. Phase 1: Claude builds a plain HTML preview with contenteditable text areas for fast, in-browser editing. Phase 2: the user iterates on content and design by editing directly in the browser and pasting changes back. Phase 3: once content is locked, Claude converts the HTML to MJML, a markup language that compiles into…
Claude Code can generate interactive HTML pages that serve as workflow interfaces for batch decision-making. Instead of answering questions one at a time in the terminal, users ask Claude to build an HTML file with one card per question, text inputs for responses, and a “copy to clipboard” button at the end. The user opens the file in a browser, answers each question, copies all…
When a CLI tool’s documentation and config file don’t have the option you need, you can read the compiled binary with the Unix strings command. strings /path/to/binary | grep toml extracts readable text embedded in the binary, including config field names and struct tags. This technique revealed an undocumented args field in agent-deck’s config that accepts startup flags like…
Ghostwriting LinkedIn posts with Claude Code follows a three-phase workflow: research (pull the subject’s public presence), voice analysis (study their actual posting patterns for length, structure, and tone), and drafting (combine source material with voice data). The AI draft is intentionally imperfect. The subject’s reaction to it, editing lines and saying “I’d never say…
A Claude Code “Chief of Staff” workflow gives Claude persistent memory across sessions by using a custom slash command that reads a folder of markdown notes before each conversation begins. The user creates a file at ~/.claude/commands/chief-of-staff.md with instructions, then maintains a “brain dump” file with project statuses, open loops, and client notes. Running…
On a Mac, Keyboard Maestro can be set up to launch Claude Code with a single keyboard shortcut. The macro uses an AppleScript action that opens a new iTerm2 tab and types claude into it. The working approach uses write text rather than passing the command directly, because the direct method bypasses the shell profile and fails to find the claude executable. Terminal.app users can use the simpler…
A Claude Code skill is a custom slash command defined by a markdown file in ~/.claude/skills/ . Each skill lives in its own folder with a SKILL.md file containing YAML frontmatter (name, description, allowed tools) and plain-English instructions. When invoked, Claude loads the file and enters a workspace shaped by those instructions. Skills support arguments via $ARGUMENTS , live data injection…
Deleting a sensitive file from a GitHub repo and committing again does not remove it from history. Git stores every commit permanently, so anyone can retrieve old files from the commit log. BFG Repo-Cleaner rewrites git history to strip files from every commit. The process involves creating a mirror clone, running BFG to remove target files, then force-pushing the clean version. After cleaning,…
Claude Code can automatically rename batches of images with descriptive, searchable filenames by visually analyzing each photo. Users point Claude at a folder of images and ask it to rename them based on what it sees. Because Claude Code is multimodal, it reads the visual content of each image and generates descriptive names like 001_mom-working-from-home-laptop-on-kitchen-table.jpg . The process…
Claude Code can automate Instagram competitive research by scraping competitor profiles via Apify, downloading images locally, and building a swipe-style review app as a local HTML page. The app displays each post with its image, caption, like count, and comment count. Users vote yes/no on each post, and results are saved in the browser’s localStorage. Patterns in engagement, content type,…
The /spawn skill in Claude Code opens a new terminal window with a fresh Claude session pre-loaded with context from the current one. It works by writing a context brief to /tmp/spawn-brief.md and a launcher script to /tmp/spawn-launch.sh , then using AppleScript to open a new iTerm2 window that runs the launcher. The file-based approach avoids the bash quoting problems that break when trying to…
Meeting transcripts from tools like Granola, Otter.ai, or Fireflies can be fed directly to Claude Code along with supporting context like emails and Slack messages. Claude Code reads the raw transcript, extracts key discussion points, and drafts a structured proposal in about 15 minutes across three iteration rounds. The AI cross-references actual source material rather than relying on memory,…
When multiple LinkedIn scrapers all return empty results, the problem is likely the input, not the tools. Seven Apify scrapers returned nothing for 45 minutes because the LinkedIn username had a hyphen ( alex-dobrenko ) that was omitted ( alexdobrenko ). Most scrapers return silent failures (empty arrays, no error message) for non-existent profiles. The debugging principle: when N different tools…
Skills and agents are Claude Code’s two extension mechanisms. Skills (also called commands) are invoked by the user with a slash command like /chief-of-staff and create an interactive conversation environment. Agents are invoked by Claude itself to handle focused, single-purpose tasks like code review or research. Skills live in .claude/skills/ , agents in .claude/agents/ . Skills are rooms…
Supabase SQL queries can be run directly from the terminal using the Management API and a curl command, bypassing the web dashboard. The Supabase CLI stores an access token in the macOS Keychain after supabase login . That token, extracted with security find-generic-password and decoded from base64, authenticates curl POST requests to the Management API’s database query endpoint. This…
tmux (terminal multiplexer) splits one terminal into multiple independent sessions that keep running in the background. For Claude Code users, this means running 3-4 Claude sessions in parallel, each on a different project, and switching between them in under a second. Sessions persist when detached, so Claude keeps working while you check other sessions. Key commands: tmux new -s name to create,…
A unit test is a small piece of code that checks whether another piece of code works correctly. Claude Code can generate unit tests for an existing project by reading source files, identifying pure functions (functions with inputs and outputs but no side effects), and writing Vitest test cases for each. In one real project, Claude Code wrote 52 tests in 30 minutes covering validation, parsing, and…
Claude Code’s memory system relies on three files: CLAUDE.md (behavioral rules), MEMORY.md (current project state), and learnings.md (corrections and preferences). MEMORY.md has a hard cap of 200 lines, and content beyond that is silently dropped. Splitting a monolith CLAUDE.md into modular rule files under ~/.claude/rules/ improves rule compliance from 92% to 96%. Adding an…
CLAUDE.md is the instruction file Claude Code reads at the start of every session. It lives at ~/.claude/CLAUDE.md (global) or ./CLAUDE.md (per project). Oversized CLAUDE.md files waste tokens and slow responses. The fix is to keep only behavioral rules in CLAUDE.md and move reference documentation to ~/.claude/references/ files that Claude reads on demand. A real setup reduced 1,578 lines to 107…
My friend Jordan recommended the idea of a got-got file and that’s literally what it is. A file that captures every time you get got. What does it mean to get got? It means that you did something based on something you read or saw or experienced, like a headline that made you click, or an ad, or a book title, or anything that really made you go deeper and sometimes buy the thing.
So there’s this article from StrongDM about how they build software with AI that boils down to No Humans Allowed. They call it “Grown code” which makes me think of Soylent for some reason, aka it’s dystopian and fascinating in equal measure, but most importantly: it’s maybe a sign of what’s to come for non-coding. There’s a few ideas in here that are worth…
The post describes how I developed an efficient workflow for creating presentation slides using Claude Code. I initially attempted PowerPoint slides, which proved cumbersome with multiple text blocks and difficult repositioning. After discovering HTML-based slides worked better, I realized I could streamline the process further. My breakthrough came from simply asking Claude to create online…
Boris Cherny - the person who literally built Claude Code - just posted his 10 tips for using it. I read through them nodding the entire time. I’m a comedian. I don’t write code. Well - I didn’t used to. Now I kind of do? But the point is: I started using Claude Code for non-coding work, and I apparently arrived at the same workflows as the guy who made the thing.
Okay, this video is all over the place. I’m editing it right now. It’s chaos. So I’m going to show you real fast, a speedy version of the whole thing, which is going from voice vomit, which I’m doing right now this in Super Whisper . That’s what this is all the way to publishing something on my blog. And so we’ll do that real fast. So let me have an idea here.…
I used Claude Code to rapidly build a timezone picker for a signup form. What used to require months of learning or thousands in developer fees now takes about ten minutes. The magic isn’t just speed - it’s the ability to create personalized solutions that match your exact vision. The Problem I needed a timezone converter for my cohort signup form, as international participants…
So I just got off a call with someone interested in learning all the claude code stuff. It was fun and cool and i muttered a lot, laughed too loud, and was generally anxious about sounding smart or making myself look like anything other than an idiot. Healthy stuff! But, and, yet, the one thing i kept saying as an answer to this kind friend’s questions was simply this: just ask claude Not…
So here’s what I did. For building the actual email first, How to Build Emails with Claude Code and MJML covers that part. 1. First Stop: Mail-Tester I went to mail-tester.com , grabbed the weird little temporary email address they gave me, added it to my Loops list, and sent my email. Score came back great. 9.5/10. Huge win. 2. Then I Tried My Personal Gmail I added my own personal email to the…