news
Skill0 starsChanged 5 months agoMITView on GitHub
Pull a fresh AI & tech news digest from across the web — model releases, capability jumps, pricing changes, and practical "how to use it" updates. Invoke when the user types /news, asks "what's new in AI", "what did OpenAI/Anthropic/Google ship", "what should I be reading this week". Supports /news today, /news week, /news month, and free-form topics like /news agents.
10 other projects use a version of this.
Adopted by
Most recently taken up by
- lowzj/news-skilltheir own version, first seen Nov 15, 2026
- CKBrennan/overtone-news-mcptheir own version, first seen Nov 15, 2026
- galiprandi/job-seekertheir own version, first seen Nov 14, 2026
- xiaoyangtx996/Searching-skilltheir own version, first seen Nov 8, 2026
- Yafan-Yang/ai-tradingtheir own version, first seen Jul 5, 2026
- carlvellotti/codex-for-pms-skills-bonustheir own version, first seen Jun 12, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- galiprandi/job-seeker★ 136same name
- carlvellotti/codex-for-pms-skills-bonus★ 8same name
- Yafan-Yang/ai-trading★ 4same name
- CKBrennan/overtone-news-mcp★ 2same name
- Ahmadjamil888/imos★ 1same name
- lowzj/news-skill★ 1same name
- shristi2211/Claude_Skills★ 1same name
- xiaoyangtx996/Searching-skill★ 1same name
- nbirkner/digital-brain★ 0same name
- Olayeeeyo/claude-skills★ 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 repository2 other files it gives its agents.
- brand-voiceSkill
- meeting-prepSkill
The file
We split this skill into the rules it gives an agent
9 rules. 0 are common, 8 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
8
Distinctive
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
- Distinctive: line 14
Default tone: terse, scannable, link-rich.
- Distinctive: line 34
Parse args: Default window = 7 days.
- Distinctive: line 57
Keep it terse. No filler.
- Distinctive: line 82
Action shortlist: only appears if there's at least one ★ item.
- Distinctive: line 88
Don't paraphrase from training data — this is *news*, time-bound.
- Distinctive: line 89
Don't include items older than the window unless asked.
- Distinctive: line 90
Don't summarize anything you couldn't actually fetch.
- Distinctive: line 91
Don't editorialize or lecture.
- Very specific: line 97
If you say "also check X" during a run, add it to the right bucket for this run and offer to persist it to sources.md for next time.
“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
News Skill — AI & Tech Digest
Replaces the "open 12 tabs every morning" habit with one command: a fast, scannable, link-rich brief on what actually shipped.
Customize this skill — this is where the expertise lives. A generic source list gives generic results. Edit the Audience line below and the source list in
sources.mdto match your field: the outlets you trust, the people you follow, and the tools/topics you want flagged. The specifics are what turn a generalist into an expert.
Audience (edit me): someone who tracks AI and wants the signal without the noise — model releases, pricing, and practical how-to.
Default tone: terse, scannable, link-rich. No preamble, no fluff. The reader skims.
Invocation
| Input | Behavior |
/news | Full digest, last 7 days |
/news today | Last 24 hours only |
/news week | Last 7 days (default) |
/news month | Last 30 days |
/news <topic> | Free-form topic (e.g. /news agents, /news Anthropic, /news open models) |
Combine freely: /news today agents.
Execution flow
- Parse args. Default window = 7 days.
- Fan out parallel searches. Issue all WebSearch / WebFetch calls in a single message — never serial. Aim for 12–15 parallel calls per run.
- Dedupe by headline / URL similarity.
- Tag each item with: source, date, category, and a ★ flag if it's directly relevant to your stack (the list you define in
sources.md). - Synthesize into the output format below.
Source buckets to hit in parallel live in sources.md — read it when planning the fan-out. At minimum:
- Official release pages: the labs you care about (Anthropic, OpenAI, Google, etc.).
- Practical / how-to trackers: independent writers who surface what's actually useful.
- Search:
"<lab>" release announcement <date>and"<topic>" technique <date>.
★ Relevance tagging
Flag an item with ★ if it touches something on your watch-list (defined in sources.md) — a tool you use, a platform you build on, a competitor, a topic you're tracking. For each ★ item, add a one-line What to do: (e.g. "Test the new model on our eval set", "Update our MCP config for the new transport").
This filter is the expertise. It's the difference between "here's all the news" and "here's what matters to you."
Output format
Keep it terse. No filler.
# News digest — <window> (<today's date>)
## AI — releases
- **<Headline>** — <1–2 sentence summary>. [Source](url) · <YYYY-MM-DD>
## AI — practical / how-to
- **<Headline>** — <what it is, why it's useful>. [Source](url) · <YYYY-MM-DD>
## <Your custom category>
- ★ **<Headline>** — <summary>. [Source](url) · <YYYY-MM-DD>
**What to do:** <one line>
---
**Action shortlist** (★ items only):
1. <action>Rules:
- Max 5 items per section. More than that, pick the highest-signal ones.
- No item without a working source link. Can't link it → drop it.
- Date every item. Can't verify the date → mark
· (date unclear)and demote. - Lead with ★ items in each section.
- Action shortlist only appears if there's at least one ★ item.
What NOT to do
- Don't paraphrase from training data — this is news, time-bound. Every claim needs a recent web source.
- Don't include items older than the window unless asked.
- Don't summarize anything you couldn't actually fetch. If the snippet is too thin, drop it rather than guess.
- Don't editorialize or lecture. State what happened, link the source.
Extending the source list
If you say "also check X" during a run, add it to the right bucket for this run and offer to persist it to sources.md for next time.
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.

