newsletter-ai-skill
CLAUDE.md0 starsChanged 8 days agoMITView on GitHub
This repo holds the /newsletter-ai Claude Code skill, the scripts that publish the newsletter, and the Hugo site it publishes to. Scheduled runs don't load this file. scripts/weekly.sh passes --restricted, which skips the project CLAUDE.md (verified in docs/specs/2026-09-13-weekly-local-publishing.md). Anything that has to reach a scheduled run belongs in .claude/skills/newsletter-ai/SKILL.md, not here. scripts/weekly.sh commits to main, pushes to a public site and spends roughly $7–10 of the user's subscription quota (capped at $10). Never run it on your own initiative, and…
In this repositoryOne other file it gives its agents.
- newsletter-aiSkill
The file
We split this CLAUDE.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 22
Run
make check: after changing the skill, the checker or the scripts: checker unit tests, an offline test of weekly.sh against a fake …
≈ “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
newsletter-ai — Project Guidelines
This repo holds the /newsletter-ai Claude Code skill, the scripts that publish the newsletter, and the Hugo site it publishes to.
Scheduled runs don't load this file. scripts/weekly.sh passes --restricted, which skips the project CLAUDE.md (verified in docs/specs/2026-09-13-weekly-local-publishing.md). Anything that has to reach a scheduled run belongs in .claude/skills/newsletter-ai/SKILL.md, not here.
Don't publish
scripts/weekly.sh commits to main, pushes to a public site and spends roughly $7–10 of the user's subscription quota (capped at $10). Never run it on your own initiative, and don't commit or push an issue yourself. If the user asks you to publish, say what the command will do and confirm before running it.
The runbook — scheduling, held issues, logs, publishing by hand — is docs/customising.md → Scheduled publishing.
Previews jam the scheduler
/newsletter-ai web:./site writes site/content/posts/<today>.md, the same file a scheduled run creates, and the wrapper refuses to run while the tree is dirty. Delete a preview post once the user has finished with it, or send it outside the repo with web:~/preview-site, which gives the markdown without a rendered site.
When you change something here
- Never edit
site/themes/PaperMod/. It's vendored at the commit pinned in.papermod-sha; update it the waysite/README.mddescribes. - After editing
scripts/weekly.sh, tell the user to re-runscripts/install-agent.sh. launchd runs an installed copy, so the repo's version has no effect until they do. - Add a line to
CHANGELOG.mdunder today's date for any change a user or forker would notice, and bumpversionin.claude/.claude-plugin/plugin.jsonwhen the change touches.claude/(installers are pinned to that version). Published issues go under Issues. - Run
make checkafter changing the skill, the checker or the scripts: checker unit tests, an offline test ofweekly.shagainst a fakeclaude, and lint. - If
check_issue.pyrejects a post, fix the post. The checker is the only thing standing between a bad issue and the live site; loosening a rule to get an issue out is never the answer.
Running /newsletter-ai
The run rules live in .claude/skills/newsletter-ai/SKILL.md under "Run rules". They govern the run, and take precedence over general guidance about fanning work out to subagents or tracking it as tasks.
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.

