youtube-skills
CLAUDE.md0 starsChanged 3 months agoMITView on GitHub
Agent Skills repo for YouTube. Each directory under skills/ must remain independently installable because npx skills add owner/repo@skill copies only that skill. The first release intentionally keeps its runtime inside the only skill. Do not add a root package or make the skill import files outside its own directory. If a second skill later shares enough runtime to justify a package and fan-out step, add an explicit sync check like reddit-skills. - plan and publish without --yes are strict local…
In this repository2 other files it gives its agents.
- AGENTS.mdAGENTS.md
- youtube-shorts-publishSkill
The file
We split this CLAUDE.md into the rules it gives an agent
6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
All writes are scoped to the user's own single channel.
- Distinctive: line 33
Never infer Made for Kids or realistic synthetic-media declarations.
- Distinctive: line 34
Never claim API-confirmed Shorts classification.
- Distinctive: line 36
Verify a write by the returned video ID.
- Distinctive: line 40
Never commit OAuth client JSON, authorization codes, tokens, upload session URLs, operator media, or audit/state files.
- Distinctive: line 46
Run the root-only report and proof contract checks from the repository root:
≈ “Common” is an estimate: at least 1 in 5 of the 28,029 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
youtube-skills
Agent Skills repo for YouTube. Each directory under skills/ must remain independently
installable because npx skills add owner/repo@skill copies only that skill.
Layout
skills/youtube-shorts-publish/
├── SKILL.md
├── agents/openai.yaml
├── package.json
├── scripts/ self-contained zero-dependency Node CLI and tests
└── references/ dated platform/API evidence and setup guidance
skills.sh.jsonThe first release intentionally keeps its runtime inside the only skill. Do not add a root package
or make the skill import files outside its own directory. If a second skill later shares enough
runtime to justify a package and fan-out step, add an explicit sync check like reddit-skills.
Rules
planandpublishwithout--yesare strict local dry runs: no OAuth reads, token refresh, network access, config directories, or audit writes.- A real write requires per-invocation
--yesplus the reviewed--expected-plan-id; default privacy isprivate; subscriber notifications default to false. - Each plan ID is one-shot by default. Snapshot reviewed bytes and atomically persist the attempt
before mutation; never auto-retry an ambiguous operation or add
--new-attemptwithout explicit user authorization after a Studio check. - All writes are scoped to the user's own single channel. No multi-account orchestration, proxy rotation, quota evasion, or enforcement bypasses.
- Never infer Made for Kids or realistic synthetic-media declarations. Both are explicit inputs.
- Never claim API-confirmed Shorts classification. The API can prove upload, processing, metadata, and visibility, not Shorts Feed placement.
- Verify a write by the returned video ID. Do not search by title, auto-delete, or re-upload after an ambiguous or still-processing result.
- Platform claims belong in
references/, with the verification date and official source. Mark absence-of-field conclusions as inference. - Never commit OAuth client JSON, authorization codes, tokens, upload session URLs, operator media, or audit/state files.
- stdout stays structured JSON; human progress goes to stderr; secrets belong in neither.
Validation
Run the root-only report and proof contract checks from the repository root:
ruby .github/scripts/check-zero-publish-report.rb
ruby .github/scripts/check-zero-publish-proof.rbThen run from the skill directory:
npm run check
npm testThen run the system skill validator from the repo root when available. CI also installs the skill through the real skills CLI and exercises the installed copy.
Publishing
This skill repository is public, like L4A-ai/reddit-skills, because skills.sh resolves and
installs it directly from GitHub. Pushing main releases the skill. There is no separate submission
API: the skills.sh listing appears from anonymous npx skills add L4A-ai/youtube-skills telemetry.
CLAUDE.md and AGENTS.md here are byte-identical twins. Edit both or neither.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

