← Overview
GitHub Trending · Saturday

14 November 2026

12 repos rising that day. Updated

git diffchanged filesbundlerno LLM, rulesLLM agentreads, searchesocr scanwhole filescommentsline-level
+215 that day★ 50.7KGo

AI code review with deterministic pipelines and an LLM agent

Open Code Review is a CLI for AI code review that began as Alibaba's internal assistant. It reads Git diffs and an LLM agent writes line-level comments.

  • File selection and bundling are handled by engineering logic, not the model, so no important change is missed. The agent can read full files and search the codebase.
  • ocr scan reviews whole files to audit a codebase with no meaningful diff. A built-in ruleset covers NPE, thread-safety, XSS, and SQL injection.
  • In the repo's own benchmark it uses about 1/9 of the tokens of Claude Code on the same model, with higher precision and F1 but lower recall.
authorwrite guidescalibratecompat dataevaluateeval dashboardagentcoding agentskillsmodern weblegacy-scanfind old code
+1.9K that day★ 4.9KTypeScript

Source repo of modern web guidance for coding agents

This repo holds the code and docs for authoring, calibrating, and evaluating Modern Web Guidance, a set of web best-practice skills for coding agents.

  • The skills bring web platform expertise and browser compatibility data to agents, steering them away from legacy patterns toward modern web features.
  • Without installing, search, retrieve, and legacy-scan run through npx modern-web-guidance@latest. An Eval Dashboard shows evaluation results.
  • Stewarded together by the Chrome, Edge, Firefox, and MDN teams and the web development community.
trynpx modern-web-guidance@latest install
cs scanrepo or difffindingsdeduplicatedvalidatecheck each oneSDK + Actionautomate in CIexportSARIF, CSVpatchverify the fix
+751 that day★ 13.2KTypeScript

CLI and SDK to find and fix security vulnerabilities

Codex Security helps find, investigate, and fix vulnerabilities in code, through the Codex plugin, the @openai/codex-security CLI, a TypeScript SDK, or a GitHub Action.

  • Scans can cover a whole repo, selected paths, or Git changes. Deep mode repeats Standard scans and returns their deduplicated findings.
  • Candidate findings can be validated, then patched, or an existing fix can be verified. Results export to SARIF, JSON, or CSV.
  • Findings can be tracked in Linear, Jira, or GitHub, and third-party scanner results can be previewed and saved locally.
trynpm install --global @openai/codex-security
market datamulti-sourceLLManalysisdashboarddecision reportnewsreal-timescheduleActions, Dockerpushchat, email
+256 that day★ 71.7KPython

Multi-market stock analysis with an LLM and automated alerts

An LLM-based stock analysis system that combines multi-source market data and news, builds a decision dashboard report, and sends it automatically. Its main README is in Chinese.

  • It covers A-share, Hong Kong, US, Japan, Korea, and Taiwan stocks plus ETFs, with price data, K-lines, technical indicators, news, and announcements.
  • Reports go to WeCom, Feishu, Telegram, Discord, Slack, or email, and can be scheduled through GitHub Actions, Docker, or local jobs.
  • There is a web and desktop workbench, plus agent Q&A with 15 built-in strategies. The README states this is not investment advice.
Maybearchived 2023Surecommunity forkyour serverDockerclientsweb, desktopAPIscripts, agents
+925 that day★ 14.8KRuby

Personal finance app, a community fork of Maybe Finance

Sure is a fully working personal finance app that can be self-hosted with Docker. This repo is a community-maintained fork of the abandoned Maybe Finance project.

  • The Maybe Finance team stopped developing its app in mid-2023 and open-sourced the code. Sure continues that codebase.
  • It can be reached from a browser, the macOS desktop app, the mobile app, API clients, and LLM agents.
  • For local development there are bin/setup, bin/dev, and demo data through rake demo_data:default.
trybin/setup && bin/dev
agentplan or diffPlannotatorlocal browseryoumark upagentreviseshooksends backfeedbackone click
+108 that day★ 10.5KTypeScript

Browser review space for plans and diffs from AI agents

Plannotator is a local, browser-based review surface for coding agents. You mark up plans, specs, HTML, or diffs, then send the feedback back to the agent.

  • It connects to the agent through plugins, hooks, and commands. Supported agents include Claude Code, Codex, Copilot CLI, Gemini CLI, OpenCode, and Kiro.
  • Code review works on local changes or remote PRs, with Git, GitButler, Jujutsu, Perforce, GitHub, and GitLab.
  • Plannotator Inbox gathers decisions, prototypes, and plans from many agents in one place.
tryplannotator inbox
checksHTTP, TCP, DNSUptime Kumaevery 20 salertsSlack, emailstatus pageown domain90+ servicesnotifications
+71 that day★ 93.9KJavaScript

Self-hosted monitoring tool with many check types

Uptime Kuma is an easy-to-use self-hosted monitoring tool. It watches service uptime and sends notifications through many channels.

  • It can monitor HTTP(s), TCP, keyword, JSON query, WebSocket, Ping, DNS, Push, Steam game servers, and Docker containers at 20-second intervals.
  • Notifications go through Telegram, Discord, Gotify, Slack, Pushover, SMTP email, and 90+ other services.
  • It offers multiple status pages mappable to domains, a ping chart, certificate info, proxy support, and 2FA.
trydocker compose up -d
memberspeople, agentsrelaysigned eventsroomschannelsclientsdesktop, mobilegit, reviewpatchesworkflowsrun in a room
+97 that day★ 37.3KRust

A workspace where humans and agents build on your own relay

Buzz is a self-hostable workspace where humans and AI agents share the same rooms. It is a Nostr relay: every message and action is a signed event in one log.

  • Agents get their own keypair, channel memberships, and audit trail, with the same affordances as a human teammate.
  • Agents can open repos, send patches, review code, run workflows, edit canvases, and join voice huddles.
  • The relay and mobile client live in this repo and the desktop app in block/buzz-app. Licensed Apache 2.0.
skillsClaude Codemarket reviewcharts, newstrade planrisk checksnext roundrepeatimprovereview habitsjournallog trades
+69 that day★ 4.2KPython

Claude Skills for the workflow of investors and traders

A Claude Skills-based trading workflow toolkit for time-constrained individual investors, covering market analysis, technical charting, economic calendars, screeners, and strategy development.

  • The goal is to structure the process: market review, risk management, trade planning, journaling, and continuous improvement, not to hand buy or sell decisions to AI.
  • It is designed for long-term investing, ETFs, and dividend stocks as the core, with disciplined swing trading as a satellite.
  • The repo states it is not financial advice and not a signal service. MIT license.
listenertailcataddressyou share itclienttailcat + addrDERPbootstrap relayNAT traversalmagicsockdirect UDPWireGuard
+256 that day★ 9.5KGo

netcat over Tailscale's data plane, no control plane

Tailcat acts like netcat but over Tailscale's data plane, with WireGuard-encrypted tunnels between two machines. Connection metadata is exchanged out of band, however you like.

  • One side runs tailcat as a listener and gets a short address. The other side passes that address to the client to connect.
  • The first connection goes through a DERP server, then NAT traversal tries to upgrade to direct UDP. DERP stays the relay of last resort.
  • No Tailscale account or root access is needed, and routing and DNS are untouched. It ships as a CLI and a Go library.
trytailcat
sourcevideo, captureGPU pipelinereal timescreenenhanced viewdenoiseHDRsuper resframe genrecordingpicture + audio
+256 that day★ 1.6KC++

Windows video player with real-time GPU enhancement

Veyra is a video player and capture tool for Windows that enhances the picture in real time on the GPU, for local video, images, capture cards, and game streaming.

  • Main features are Super Resolution, denoising, HDR, and Frame Generation, with PS5, Xbox, and PC streaming sources supported.
  • Written in C++ and DirectX 12 for Windows 11 with NVIDIA and AMD GPUs. GPL-3.0 licensed.
  • The README reports 79K+ portable-package downloads as of 2026-11-08, counted as download events, not unique users.
agentClaude, CursorMCP serveror CLIChromelive browsertracesperformancepuppeteerautomationconsolenetwork, shots
+67 that day★ 55.5KTypeScript

Chrome DevTools for coding agents through MCP and a CLI

chrome-devtools-mcp lets a coding agent control and inspect a live Chrome browser, as an MCP server or as a CLI.

  • It records performance traces through Chrome DevTools, inspects network requests, takes screenshots, and reads console messages with source-mapped stack traces.
  • Automation uses puppeteer and waits for action results automatically.
  • Browser content is exposed to MCP clients, so avoid sensitive data. Google Chrome and Chrome for Testing are officially supported.
trynpm i chrome-devtools-mcp@latest -g