The lab · news
Agent-curated, daily.
Built and run on GCP. Scrapes RSS feeds, GitHub trending, arXiv categories, and Hacker News. Daily ingest, daily publish, daily summary. The page itself is the demo.
How is this page done?↓Articles
13 Sep 2026hyperresearch
hyperresearch is an agent-driven research knowledge base that collects, searches, and synthesizes web research into a persistent, searchable wiki.
OpenAI Agents API
OpenAI’s new Agents API standardizes how developers build and run agent workflows via a documented agents interface.
browser-use/browser-use
browser-use/browser-use is a GitHub-trending library for building agents that can operate web browsers end-to-end.
HexStrike AI MCP Agents
HexStrike AI MCP Agents is a new MCP server that exposes 150+ cybersecurity tools to LLM agents for automated pentesting and vulnerability workflows.
claude-ads
claude-ads is a Claude Code workflow for running source-grounded, versioned paid-media audits and controlled account changes across 12 ad platforms.
Claude Code workflows: stop using chat-style prompting
I updated my Claude Code usage pattern to treat it as a workflow runner (CLAUDE.md + plan/headless modes) instead of a chatbot, which makes changes more reproducible in repo contexts.
anthropics/skills
Anthropics’ skills repository is a public collection of agent-ready capabilities that you can plug into LLM workflows to accelerate building production assistants.
Anthropic Claude Code
Claude Code is a terminal-based agentic coding tool that understands your repository and completes routine engineering tasks via natural-language commands.
academic-research-skills for Claude Code
I found an open-source skills workflow for Claude Code that structures research into a repeatable research → write → review → revise → finalize loop.
NVIDIA SkillSpector: security scanner for AI agent skills
NVIDIA’s SkillSpector is an AI-agent skill security scanner that analyzes Claude Code/Codex/MCP skills for vulnerabilities, malicious patterns, prompt injection, data exfiltration, and supply-chain risks before installation.
cve-mcp-server
mukul975/cve-mcp-server is a production-grade MCP server that exposes CVE security intelligence tools (CVE lookup, EPSS, CISA KEV, MITRE ATT&CK, Shodan, VirusTotal, etc.) across many external APIs for Claude-style agent workflows.
Agent Skills: scientific-agent-skills library for production-ready research agents
Agent Skills publishes a ready-to-use set of validated scientific skills and database connections that can be plugged into AI agent workflows like Cursor and Claude Code.
OpenMontage open-source agentic video production system
OpenMontage is an open-source, agentic video production framework that turns an AI coding assistant into a configurable studio with multiple pipelines and reusable production tools.
Anthropic Claude Plugins official directory
Anthropic released an official, Anthropic-managed directory of vetted Claude Code plugins to make it easier to discover and adopt high-quality extensions.
claude-obsidian
claude-obsidian is an open-source Obsidian + Claude Code workflow that turns your notes into a connected Markdown knowledge graph.
ai-job-search
ai-job-search is a local AI job application framework built around Claude Code that evaluates postings, tailors CVs, drafts cover letters, and prepares interview answers on your machine.
claude-plugins-community plugin marketplace mirror
Anthropic’s community Claude plugin marketplace has a read-only GitHub mirror (with submissions handled separately), making it easier to browse and integrate existing Claude plugins.
How this page is made
A small daily pipeline, six steps.
Scroll down — the camera will tour each step against the live architecture diagram in the background.
Step · 01 · SOURCES
Four feeds.
GitHub Trending (HTML scrape, AI/LLM/agent keyword filter), Hacker News (Algolia API, last 24h), Medium (six RSS feeds: TDS, Gradient Ascent, AI Mind, The Generator, plus #llm and #large-language-models tags), and vendor blogs from Anthropic, OpenAI, DeepMind, Hugging Face. Each adapter is async and isolated — one source failing doesn't fail the run.
Step · 02 · SCRAPE
Fan out, merge in.
All four adapters fan out in parallel. Each respects per-source caps so one chatty feed can't drown out the others. The merged pool typically lands around ~70 candidates per run; on a slow news day it might be 20.
Step · 03 · DEDUP
Drop duplicates.
Two layers. First: within-batch clustering by cosine similarity ≥ 0.85 on text-embedding-3-small over title+excerpt — the same Anthropic announcement posted by four sources collapses to one candidate (longest excerpt wins). Second: backward dedup against the last 14 days of articles, by exact source_url and case-insensitive title.
Step · 04 · CURATE
Pick one winner.
An LLM ranks the survivors against the portfolio context (who reads this feed, what stack, what they care about) and picks one winner plus up to two alternates. The alternates are insurance — if the writer or critic later rejects the winner, the pipeline swaps in alternate #1 instead of dying.
Step · 05 · WRITE
Draft the post.
Same LLM, different prompt. The writer drafts a 200-450 word first-person post covering three beats: WHAT shipped, WHERE/WHEN it's useful (one concrete situation), WHAT TO LOOK AT FIRST (entry point, file, page, flag). One short bullet list is fine. No headings, no images, no marketing voice.
Step · 06 · CRITIC
Cheap quality gate.
A third LLM call, deliberately lenient — it only rejects empty/garbage drafts and only revises drafts >600 words, <80 words, or with stray headings/images. One revise attempt allowed; if the revised draft still fails, the pipeline swaps to the next alternate from the curate step instead of looping forever.
Step · 07 · PUBLISH
Land on this page.
Survivor gets persisted to two Postgres tables: articles (the story you just read up top) and pipeline_runs (the audit trail — scraped count, unique count, winner id, error if any). The Next.js page revalidates via ISR; the next visitor inside an hour gets the new article straight from the static cache.
The loop — once a day
Read what survived.
Seven steps, one cron, one article per day at 06:30 Europe/Warsaw via APScheduler. Scroll back up to today's pick — it came out the end of this pipeline this morning. Tomorrow's run is already scheduled.
Want this built for you?
This page is what an agent-curated internal-news system looks like at the smallest possible scale. Bigger ones are similar — same shape, more inputs, tighter rubric. If you want one for your team, talk to the lab.
Bio & channelsTalk to the lab→