newsletter-ai
Skill4 starsChanged 3 days agoMITView on GitHub
Generates the weekly executive AI intelligence newsletter: researches the last 7 days of AI developments (with 30-90 day context where needed) and writes them up as a 15-part executive brief — the Top 20 developments, model intelligence, open-vs-closed, talent, funding/M&A, production deployment, agents, chips/compute, data centers/energy, the global AI race, a company watchlist, key numbers, under-the-radar signals, what changed this week, and a closing synthesis. Defaults to markdown chat output; pass -o/--output yaml (or --yaml) to instead produce a sourced YAML document, and add -r/--report path/to/folder-or-report.html to render it straight into an interactive, light/dark, multi-tab HTML report (with per-tab References accordions, its source YAML fused in for single-file distribution, and a Download-data button) written next to a copy of the YAML. Use when the user runs /newsletter-ai (aliased as the shorthand /nltr-ai — treat both identically), or asks to draft, write, or update this week's AI newsletter/intelligence brief.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- ChrisAdkin8/newsletter-ai-skilltheir own version, first seen Nov 9, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- ChrisAdkin8/newsletter-ai-skill★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository58 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- anywidget-generatorSkill
- brandkitSkill
- design-taste-frontendSkill
- diagram-designSkill
- docxSkill
- frontend-designSkill
- full-output-enforcementSkill
- gpt-tasteSkill
- grill-meSkill
- high-end-visual-designSkill
- imagegen-frontend-mobileSkill
- imagegen-frontend-webSkill
- image-to-codeSkill
- impeccableSkill
- implement-paper-autoSkill
- implement-paperSkill
- industrial-brutalist-uiSkill
- jupyter-to-marimoSkill
- marimo-notebookSkill
- minimalist-uiSkill
- pdfSkill
- postrboardSkill
- pptxSkill
- redesign-existing-projectsSkill
- skill-creatorSkill
- stitch-design-tasteSkill
- streamlit-to-marimoSkill
- xlsxSkill
- brandkitSkill
- design-taste-frontendSkill
- full-output-enforcementSkill
- gpt-tasteSkill
- high-end-visual-designSkill
- imagegen-frontend-mobileSkill
- imagegen-frontend-webSkill
- image-to-codeSkill
- impeccableSkill
- industrial-brutalist-uiSkill
- minimalist-uiSkill
- redesign-existing-projectsSkill
- stitch-design-tasteSkill
- network-flowSkill
- create-lib-docsSkill
- artwallbgSkill
- commentdropSkill
- prodigySkill
- scrolls-explainerSkill
- scrolls-helpSkill
- scrolls-hideSkill
- scrolls-removeSkill
- scrolls-restoreSkill
- scrolls-setupSkill
- scrolls-unhideSkill
- scrolls-updateSkill
- stata-recipesSkill
- dynamic-promo-videoSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 1 is distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 27
Read assets/PROMPT.md in full before writing anything.
- Distinctive: line 41
Do the research it calls for: AI developments from the last 7 days, pulling in 30-90 days of prior context only where a development needs …
- Very specific: line 48
Write the newsletter following assets/PROMPT.md's structure and rules exactly, in order: the Top 20 Developments, then Parts 2-15 (Model …
- Very specific: line 54
Keep confirmed fact, company claim, reported information, and analyst inference clearly distinguished throughout, as assets/PROMPT.md …
- Very specific: line 67
yaml/yml: instead of prose, produce a single YAML document conforming exactly to the schema builder/report_data.py validates (see its … - Very specific: line 83
No
-r/--reportgiven: write the YAML to a file (or return it as a fenced code block if the user hasn't said where), and mention it can … - Very specific: line 91
-r/--report PATHgiven: run the render yourself:
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill 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 Intelligence Brief
This skill's job is to produce one edition of the weekly AI intelligence
newsletter. All of the editorial substance — role, audience, research
window, sourcing rules, the 15-part structure, editorial rules, and
writing style — lives in assets/PROMPT.md in this same directory. This
file only says how to use it; assets/PROMPT.md is the single source of
truth for what the newsletter contains and how it reads, so don't
reconstruct the brief from memory or summarize it — read the file itself.
Steps
- Read
assets/PROMPT.mdin full before writing anything. Treat it as the complete editorial brief for this edition, not a template to paraphrase. - Parse the invocation for flags:
- Output format —
-o/--outputfollowed bymd,yaml, oryml; or the shorthand--md/--yaml/--yml. No flag meansmd(the original behavior). An unrecognized value is an error — ask the user to pickmdoryamlrather than guessing. - Report path (
yaml/ymloutput only) —-r/--reportfollowed by a path: either a folder (getsreport.html+report.yamlinside it) or an explicit.html/.htmfile path (gets a.yamlsidecar with the same basename next to it). Given withmdoutput, it's a no-op — say so rather than silently ignoring it.
- Output format —
- Do the research it calls for: AI developments from the last 7 days,
pulling in 30-90 days of prior context only where a development needs
it to make sense. Use current web research and prefer the primary
sources
assets/PROMPT.mdlists (company announcements, papers, model cards, technical reports, repos, earnings, regulatory filings) over secondary reporting; cross-check important claims across multiple sources. - Write the newsletter following
assets/PROMPT.md's structure and rules exactly, in order: the Top 20 Developments, then Parts 2-15 (Model Intelligence through The Big Picture), applying the Editorial Rules and Writing Style sections throughout. Skip a section's content honestly ("No material development this week.") rather than manufacturing a story to fill it. - Keep confirmed fact, company claim, reported information, and analyst
inference clearly distinguished throughout, as
assets/PROMPT.mdrequires — don't blur them for narrative flow. - Attach a real source to every sourced claim, and never invent one.
Whenever a development traces to an actual URL you found during
research, carry that URL forward as a citation (see the YAML
sourcesfield below, or an inline link in markdown mode). When no verifiable URL exists for a claim, say so in the text instead of fabricating a link — a missing citation is honest; a fake one is not. - Produce the output in the requested format:
md(default): return the finished edition as markdown chat output, structured perassets/PROMPT.md, unless the user asks for it to be saved to a file. Cite sources as ordinary markdown links.yaml/yml: instead of prose, produce a single YAML document conforming exactly to the schemabuilder/report_data.pyvalidates (see its module docstring andvalidate()for the authoritative field list) — ametablock plus an orderedsectionslist (onekind: homelanding section, then onekind: standardsection per Part, each withitemsthat carrytitle, anicon/badgefrom the template's set, and whichever ofbody/facts/table/stats/kvfits that item's content, plussources: [{label, url}]for every claim with a real link). Useassets/templates/sample-report.yamlas a fully worked example — copy its shape rather than inventing a new one. Do not hand-number citations or write a references list yourself: list each item'ssources, and the render step below dedupes them by URL and builds the numbering and per-tab References accordions automatically, which is what guarantees every citation link actually resolves.No
-r/--reportgiven: write the YAML to a file (or return it as a fenced code block if the user hasn't said where), and mention it can be rendered into the interactive report with:python skills/newsletters/newsletter-ai/builder/build_report.py <file>.yaml <output>.html-r/--report PATHgiven: run the render yourself:python skills/newsletters/newsletter-ai/builder/build_report.py <file>.yaml -r PATHThis writes the rendered HTML and a copy of the YAML side by side (see Report path above for how PATH resolves), and by default fuses the source YAML into the HTML too (a hidden, base64-encoded
<script>blob) so the single.htmlfile remains fully reproducible even without its.yamlsidecar — the page's "Download data" button in the top bar lets a viewer pull that YAML back out. Pass--no-fuseonly if the user explicitly doesn't want the data embedded. Report the paths written back to the user.
Development
See meta/MAINTAINERS.md for this skill's layout, the YAML schema, how to
update assets/PROMPT.md or assets/templates/report.html, how to run
the test suite, and versioning conventions. It is not read as part of
carrying out a /newsletter-ai request — don't act on it while producing
an edition.
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.

