← Overview
GitHub Trending · Tuesday

17 November 2026

14 repos rising that day. Updated

agentany frameworkevent streamtyped eventsclientweb, CLIshared statetools, human
+791 that day★ 19.1KTypeScript

Open protocol connecting any AI agent to any app

AG-UI is an open, lightweight, event-based protocol that standardizes how AI agents connect to user-facing apps. The agent sends typed events over one bi-directional stream.

  • The agent emits typed events, and any client can render them, whatever framework built the agent.
  • It covers streamed tokens, tool calls, shared state, pausing for a human, and subagents.
  • The same agent can power a web app, a mobile app, a terminal, or Slack.
agentplan, diffPlannotatorannotateyoureviewfeedbackone click
+696 that day★ 12.6KTypeScript

Workspace to annotate agent plans and review code

Plannotator is a review layer between you and multi-agent work. You annotate plans, specs, and diffs, then send all feedback to the agents with one click.

  • You can annotate plans, specs, markdown, HTML, and other interactive artifacts.
  • It supports diff and PR review, plus sharing with your team.
  • It is described as working with all coding agents and personal assistants.
your appor AI agentFacade API+ eventsUniver coreSheets, DocsOffice SDKedit, calc
+1.8K that day★ 27.8KTypeScript

Open-source office SDK: Sheets and Docs for apps and agents

Univer Office SDK brings editing, calculation, collaboration, and file processing into your product for people and AI agents. It is built on an open-source core, starting with Sheets and Docs.

  • The quick start is an open-source spreadsheet editor you can connect to your application.
  • Applications and agents connect through Facade APIs and events, with clear boundaries for custom features.
  • Univer Workspace is included as a reference application for the Office SDK.
market datacharts, newsClaude skillsscreen, plantrade planrisk, journalyou decidenot the AI
+837 that day★ 6.4KPython

Claude skills for investor and trader workflows

Claude Trading Skills is a Claude Skills-based trading workflow toolkit for individual investors. It structures market review, risk management, trade planning, and journaling instead of handing buy/sell decisions to AI.

  • It covers market analysis, technical charting, economic calendars, screeners, and strategy development.
  • It is designed for long-term investors using ETFs and dividend stocks as the core, with swing trading as a satellite.
  • The project started as the author's personal effort to improve their own trading process.
youbrowser UIArcanev3Dockerhost enginecontainersimages, more
+46 that day★ 8.8KGo

Modern Docker management designed for everyone

Arcane is a modern Docker management interface. The main branch is now v3, and the last 2.x release lives on a separate branch.

  • Feature requests moved to a dedicated place, and issues are now only for bug reports.
  • Documentation covers setup, configuration, and development guides.
  • Translations are managed on Crowdin, and an SBOM is available for security transparency.
timeline403 BCE - nowmap layersterritoryplacesold to newpoetry routes18 dynasties
+288 that day★ 6.9KJavaScript

Interactive atlas of Chinese history on a timeline

Jiangshan Map is an interactive historical atlas of China from 403 BCE to the present. Scrub through time to see territories, trace ancient cities, and follow classical poetry routes.

  • It covers 18 dynasties and 198 territory snapshots.
  • It includes place-name genealogy from ancient to modern.
  • Classical poetry routes help locate cities named in history and literature.
coding agentClaude, Codexskillsdesign, buildStitch MCPserverpluginsmarketplace
+529 that day★ 9.8KTypeScript

Agent skills for Google Labs' Stitch MCP server

Stitch Skills is a library of Agent Skills for the Stitch MCP server, following the open Agent Skills standard. It works with Codex, Antigravity, Gemini CLI, Claude Code, Cursor, and OpenCode.

  • Plugins install through a marketplace, such as Codex's, or manually for other agents.
  • Plugins are split into stitch-design, stitch-build, and stitch-utilities.
  • Each skill follows the open Agent Skills standard.
topicyour querylast30dayssearch, scoresourcesReddit, X, HNsummarygrounded
+135 that day★ 68.2KPython

Agent skill that researches the last 30 days across platforms

last30days is an AI agent skill that researches a topic across Reddit, X, YouTube, HN, Polymarket, and the web, then writes a grounded summary. Results are scored by upvotes, likes, and real money rather than editors.

  • Install in Claude Code via the plugin marketplace, or with npx skills add for Codex, Cursor, Copilot, and Gemini CLI.
  • The README says the v4 pipeline is the current version.
  • The runtime skill spec lives in the SKILL file, which is the source of truth.
trynpx skills add mvanhorn/last30days-skill -g
skillor pluginquality gateslint, overlaplive evalwith vs withoutreportimpact
+873 that day★ 2.8KPython

Evaluation framework for AI agent skills and plugins

SkillEvaluator checks AI agent skills and plugins and measures a skill's effect on task results. It starts with quality checks, then compares runs with and without a skill.

  • It checks quality, security, overlap, and compatibility of skills.
  • It runs from a CLI, from CI in your repo, or wired to an agent.
  • It includes synthetic evaluation dataset generation and live agent evaluation.
trypip install skillevaluator
git diffchanged filesagent + rulestools, LLMmodelOpenAI, Claudecommentsline-level
+951 that day★ 53KGo

Alibaba's AI code review CLI, pipelines plus LLM agent

Open Code Review is an AI code review CLI that began as Alibaba's internal review assistant. It reads Git diffs, sends changed files to an LLM through an agent, and generates line-level comments.

  • Hybrid architecture: deterministic pipelines plus an LLM agent that can read full file contents.
  • It has a built-in multi-language ruleset, including NPE, thread safety, XSS, and SQL injection.
  • It is compatible with OpenAI and Anthropic endpoints.
ideas + logsraw notesagentsoutline, writeLaTeX paperwith plotslit. reviewrefine
+313 that day★ 1.6KPython

Multi-agent framework for writing AI research papers

PaperOrchestra turns ideas and experiment logs into submission-ready LaTeX manuscripts. A team of specialized agents handles outline, literature review, writing, refinement, and plotting.

  • It produces literature synthesis and visuals such as plots and conceptual diagrams.
  • It comes from Google Cloud AI Research.
  • It accepts unconstrained pre-writing materials.
coding agentClaude, CursorMCP serveror CLIChromelive browserDevToolstrace, debug
+841 that day★ 57.7KTypeScript

Chrome DevTools for coding agents via MCP

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

  • It gives access to DevTools for automation, debugging, and performance analysis.
  • It can record traces to extract performance insights.
  • It works with Antigravity, Claude, Cursor, and Copilot.
componentsdeclarativecompilerbuild stepJavaScriptsmall outputDOMsmall updates
+115 that day★ 89KJavaScript

Svelte, a compiler turning components into efficient JavaScript

Svelte is a new way to build web applications. It is a compiler that converts declarative components into efficient JavaScript that surgically updates the DOM.

  • It is MIT-licensed.
  • Development is supported by volunteers and donations through Open Collective.
  • Donated funds cover expenses such as hosting.
agentClaude, Cursorffmpeg-skill42 toolsvideoin, outlocal FFmpegno cloud
+367 that day★ 6.8KPython

Local video editor for coding agents with 42 FFmpeg tools

ffmpeg-skill gives a coding agent a local FFmpeg video editor, then the agent checks its own output. No cloud, no API keys, Python standard library only.

  • It has 42 tools with a machine-readable contract and an MCP server.
  • Features include cut, captions, reframe, loudness, and platform exports.
  • It works with Claude Code, Cursor, Codex, or any MCP client.
trynpx ffmpeg-skill