Daily radar on the newest arXiv papers about large language models, with a one-line takeaway for each.
This skill is a 3-step workflow. Everything below is the full definition — nothing hidden. Use it from the Dashboard (no code), the REST API, or an AI agent via MCP.
{
"name": "arXiv LLM radar",
"trigger": {
"type": "cron",
"everyMinutes": 1440
},
"steps": [
{
"action": "fetch",
"url": "http://export.arxiv.org/api/query?search_query=all:%22large+language+models%22&sortBy=submittedDate&sortOrder=descending&max_results=10"
},
{
"action": "ai",
"prompt": "These are arXiv paper abstracts about LLMs. For each paper give: title and a one-line takeaway in plain language. Plain text.",
"input": "{{1.body}}"
},
{
"action": "email",
"to": "you@example.com",
"subject": "arXiv LLM radar",
"body": "{{3.text}}"
}
]
}1No code — Dashboard
Hit “Use this skill” — the builder opens with every step prefilled. Edit the placeholder URLs, press run. Done.
2API — one request
curl -X POST https://api.ottersnap.com/v1/workflows \
-H "Authorization: Bearer otter_live_YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{"name":"arXiv LLM radar","trigger":{"type":"cron","everyMinutes":1440},"steps":[{"action":"fetch","url":"http://export.arxiv.org/api/query?search_query=all:%22large+language+models%22&sortBy=submittedDate&sortOrder=descending&max_results=10"},{"action":"ai","prompt":"These are arXiv paper abstracts about LLMs. For each paper give: title and a one-line takeaway in plain language. Plain text.","input":"{{1.body}}"},{"action":"email","to":"you@example.com","subject":"arXiv LLM radar","body":"{{3.text}}"}]}'3AI agent — via MCP
Glint ships an official MCP server. Paste this prompt into Claude, Cursor or Codex and it creates the skill for you:
Using the ottersnap-mcp server (npm: ottersnap-mcp), create a Glint workflow with this exact definition:
{
"name": "arXiv LLM radar",
"trigger": {
"type": "cron",
"everyMinutes": 1440
},
"steps": [
{
"action": "fetch",
"url": "http://export.arxiv.org/api/query?search_query=all:%22large+language+models%22&sortBy=submittedDate&sortOrder=descending&max_results=10"
},
{
"action": "ai",
"prompt": "These are arXiv paper abstracts about LLMs. For each paper give: title and a one-line takeaway in plain language. Plain text.",
"input": "{{1.body}}"
},
{
"action": "email",
"to": "you@example.com",
"subject": "arXiv LLM radar",
"body": "{{3.text}}"
}
]
}
Then confirm it was created and tell me how to run it once manually.fetchSTEP
http://export.arxiv.org/api/query?search_query=all:%22large+language+models%22&sortBy=submittedDate&sortOrder=descending&max_results=10
aiAI
These are arXiv paper abstracts about LLMs. For each paper give: title and a one-line takeaway in plain language. Plain text.
accepts {{1.…}} from the previous step
emailSEND
you@example.com · arXiv LLM radar
accepts {{2.…}} from the previous step
EVERY RUN = 1 CREDIT · FAILED STEPS STOP THE CHAIN · RESULTS FEED FORWARD WITH TEMPLATES
MORE SKILLS
Competitor price watch
Watch a rival's pricing page. When it changes, AI summarizes exactly what changed and emails you the delta.
Daily AI briefing
Every morning, extract your favorite news page and get a one-paragraph AI digest in your inbox. Edit the URL to your source.
Page change evidence
When a monitored page changes, capture a full screenshot archive and email yourself the link — proof of what the page looked like.