Claudemy

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites: none for the install itself: no account, no API key, no paid service, nothing to install first, and this install never asks for a key. Using the skill afterwards needs python3 with the beautifulsoup4 package (pip install beautifulsoup4) and ffmpeg. Audio rendering uses my own ElevenLabs API key when ELEVENLABS_API_KEY is set in the environment at that time (ElevenLabs is a paid service beyond its free tier); without a key, rendering falls back to the free macOS built-in voices. All of that is use time, never this install. Install "claudemy" for me. It turns a topic into a short audio course: it researches the topic on the web, maps what good sources agree and disagree on, writes a set of about 2-minute two-host audio lessons (12 by default), each with a one-screen card and one small thing to try, and packages the class with its sources cited. Repository: https://github.com/assafkip/claudemy (the skill is the whole skills/claudemy/ folder at the pinned commit: SKILL.md, references/ and scripts/) Pinned version: install the code at commit 438a6288fa74a2edad2cc86416d28308cb02ee1a exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/assafkip/claudemy into a temporary folder, check out commit 438a6288fa74a2edad2cc86416d28308cb02ee1a, and copy the skills/claudemy/ folder from that checkout. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-19: [FINGERPRINTS] - SKILL.md: SHA-256 49251c4aa514d34dcc9cde9435a08edd0e0997867b3c620d488d8fb66974863f - references/lesson-design.md: SHA-256 5492cd5b80c0ebc5375a1d805ef95efacf9fbd2b6e347befe9d3bae47ea1222c - references/lesson-example.md: SHA-256 4d22360fd8f07800733da1a17ce64ea4593922ef926c1c0449f51a1ae8c1feba - references/map-brief.md: SHA-256 08baa23fd1ff5cf73060e494ff87b0431510c4d54bceb16a9025edffc2fdc5a7 - references/reader-brief.md: SHA-256 c378243fcf0b154b78c121a6cca6778eecb9f35a4723a5537e2b964c740dd469 - references/writer-brief.md: SHA-256 ee5358fb90e4e9c20c4f9fabdce0699ac5399f86c24c72317e54b1114cb6e129 - scripts/check_lessons.py: SHA-256 e52b918ccaf9ebce24630a633bf885949711edf875fdd008ae179d15072445ce - scripts/package_class.py: SHA-256 cf628a0af958009743ca1ca09be7b2497bb280a84b5267c7bac5e42a2f4a04c0 - scripts/render_audio.py: SHA-256 b0f9bb033cf450864233a6992e8514be5e196cc564132cf02cad095efab28de4 - scripts/scrape.py: SHA-256 13e1816564ec6db128ed000edfe4a90058ac2b6113bc73337928ed76e803e615 If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md in full, skim the reference files, and read the four scripts in scripts/, then tell me in plain words: what the skill does when used, which services it connects to, whether it reads or sends any personal information or keys, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the files: - The skill folder holds 10 files: SKILL.md (the method: map the topic, build a reading list, read, agreement map, lesson plan, one sample lesson for my reaction, then the rest), five reference documents (lesson design, a worked lesson example, and the reader, map and writer briefs), and four plain readable Python scripts: scripts/scrape.py (fetches the reading list into the workspace), scripts/check_lessons.py (fails any lesson or plan that cites a page not saved on disk), scripts/render_audio.py (renders each lesson's two-voice audio) and scripts/package_class.py (builds the delivery zips and the combined cards file). Nothing is obfuscated and there is no build step. - Nothing is sent anywhere during installation itself; the only download at install time is the claudemy folder from GitHub at the pinned commit. The files name no telemetry. The upstream plugin route (Claude Code plugin "claudemy" from this repository) offers an optional configuration prompt for an ElevenLabs key stored in the OS keychain; the folder copy I am asking for has no such prompt and touches no key. - At use time, the research is web traffic I should expect: searches, then scripts/scrape.py downloading the 20 to 40 source pages listed in the class's sources.json (public pages; it skips paywalled or login-walled ones). In person mode, when I name someone to learn from, it instead fetches that person's public archive: their Substack archive API (https://<host>/api/v1/archive and /posts/<slug>), or their site's sitemap or RSS feed. Every fetched page is saved under the class workspace, and the skill's honesty rules require every cited claim to trace to a saved page, and say the saved pages are for my private study, not to republish. - The one service the lesson content itself is sent to is ElevenLabs, and only when I have set ELEVENLABS_API_KEY: scripts/render_audio.py posts each lesson's dialogue text to https://api.elevenlabs.io/v1/text-to-dialogue with my key as the xi-api-key header and receives the audio back. Without the key, rendering stays local with the macOS built-in voices. SKILL.md also says never to print the key or copy it into the workspace; if I keep it in a file, it is sourced in the same shell command as the render. - The skill writes everything for one class under ./claudemy/<topic-slug>/ in the current directory (sources, raw pages, research notes, the class with MP3s, cards and playbook, and the delivery zips). SKILL.md states the class does not read my project files, notes or settings. It makes one sample lesson first and waits for my reaction before rendering the rest. 3. When I say "yes, install", install it as a skill: copy the whole claudemy folder into the agent's skills directory, in a folder named "claudemy", keeping all 10 files. The upstream routes, the Claude Code plugin (/plugin marketplace add assafkip/claudemy, then /plugin install claudemy@claudemy) or a manual copy of the skills/claudemy folder, install the same skill; either route is fine as long as every file matches the fingerprints. 4. Verify with no network calls: frontmatter valid, all 10 files in place. 5. Report what was installed, where, and what I still need to do myself (for example: have python3, beautifulsoup4 and ffmpeg available before first use, decide myself whether to set an ElevenLabs key for natural voices, and name a topic myself; using the skill is a separate step that only starts when I ask for it, and the use-time fetches and the ElevenLabs calls disclosed above happen only then). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.