AIHOT collects items from feeds, web pages, X accounts and WeChat accounts. An LLM pre-filters them, then scores each one twice. Items that pass a threshold get a Chinese title and summary and become picks. Reports of the same story are merged into one event, and events are ranked by how many independent sources covered them in the last 48 hours.

The repo holds the website, admin panel, selection pipeline, clustering and heat code, and the full text of every prompt. Industry-specific parts live in the site/ and industry/ folders. To change topic you edit prompts, thresholds, categories and the source list, and the code stays as it is.

The author says plainly that this is an export of the live engine, not a polished generic framework, and that it was rewritten with AI help. It ships 18 public overseas AI sources as a demo. The real AIHOT source list and some AI-only features are not included. The README and site are mainly in Chinese.

Read more on interesting repos · GitHub

n8n lets you connect apps and services into automated workflows. You draw each workflow as a chain of nodes on a canvas. Each node does one job, such as fetching data, calling an API or asking a language model a question.

It also supports multi-step AI agents. Per the README, you can connect OpenAI, Anthropic, Google or open-source models and switch providers without redesigning the workflow. You can add logic, tool use and human approval steps.

Compared with purely no-code tools, n8n lets you drop into JavaScript, Python and npm packages when the visual nodes aren't enough. You can run it on your own server. The license is fair-code (Sustainable Use License), so the source is visible but it isn't a standard open-source license.

Read more on interesting repos · GitHub

Google Workspace has many separate APIs, and calling them usually means reading REST docs and hand-writing curl requests or client code. gws puts them behind one command-line tool. Every resource has --help, and you can preview a request with --dry-run before it is sent.

It does not ship a fixed list of commands. At runtime it reads Google's Discovery Service and builds the commands from that, so new API methods show up without a new release. On top of that it adds hand-written helpers prefixed with +, such as gmail +send, calendar +agenda and workflow +standup-report.

Output is structured JSON, and paginated results can be streamed as NDJSON, so it pipes easily into jq or to an LLM. The repo also includes 100+ agent skills (SKILL.md files) and a Gemini CLI extension. The README says this is not an officially supported Google product and that breaking changes are expected before v1.0.

Read more on interesting repos · GitHub

Posting the same content to many social platforms by hand takes a lot of time, especially for one-person companies and small creators. AiToEarn puts creation, publishing, engagement and monetization in one place. It covers Douyin, Xiaohongshu, Kuaishou, Bilibili, TikTok, YouTube, Instagram, X, LinkedIn and others.

You tell an agent what content you want. It calls video and image models (the README names Grok, Veo, Seedance and Nano Banana) and can make many pieces in parallel. A calendar schedules posts across platforms, and a browser plugin handles likes, follows and AI-written replies.

You can use the hosted site, plug it into any MCP-compatible assistant or OpenClaw with an API key, or self-host with Docker. Self-hosted setups can borrow the official service's platform logins through a Relay, so you don't need your own developer accounts. The README also describes a marketplace where creators take promotion tasks paid per sale, engagement or view.

Read more on interesting repos · GitHub

Agents that use real apps run into the same chores: API keys, OAuth sign-in flows, and each service's own request format. One CLI replaces that with one tool. You connect an app once, and the agent can search for the action it needs, read the docs, and run it.

Every call goes through One's hosted proxy at api.withone.ai. The proxy adds the right credentials, handles rate limits and normalizes responses, so you only deal with a connection key. This means the tool depends on One's service and an account there.

You can also chain actions across apps into reusable workflows written as JSON files. Flows support conditions, loops, parallel steps and small code modules. The init command installs an MCP server into agents such as Claude Code, Cursor and Codex, and access settings can limit what an agent may do.

Read more on interesting repos · GitHub

Awesome One is a showcase repo for One, a platform that connects apps like Gmail, Slack, Google Sheets, HubSpot, GitHub and Linear. It holds four projects you can clone, change and deploy. They rely on One's service, so you will likely need a One account to run them.

The biggest piece is Flows: more than 218 automations rebuilt from popular n8n templates. Each runs with a single one flow execute command. The README says you can paste a flow's URL into Claude Code or Cursor and it sets itself up.

The other three are fuller examples. Link is a branded page where an agency's customers connect their accounts. Inbox Run triages a Gmail mailbox, then labels, archives, logs to Sheets, adds HubSpot contacts and posts to Slack. Dev Pulse is a Next.js dashboard with live webhooks and an AI chat panel using One's MCP server.

Read more on interesting repos · GitHub