GitHub Trending — November 13, 2026
The GitHub trending list reveals an ecosystem in accelerated transition. What was once futuristic — autonomous coding agents, security embedded in the development pipeline, self-hosted monitoring — is now stable product with thousands of stars gained in days. This time, the dominance of AI coding agent tools persists, but with interesting nuances: personal finance projects, distributed communication, and LLM-driven stock analysis are all making waves.
1. Alibaba Open Code Review — The enterprise-grade code review standard
alibaba/open-code-review gained 206 stars today alone (50,474 total), consolidating itself as one of the most adopted open source tools on GitHub. But the number itself doesn't tell the most important part: this is a product Alibaba incubated internally, validated at real development scale, and is now releasing to the community under Apache 2.0.
What makes it different from other static analyzers? Its hybrid architecture. It combines deterministic pipelines with an LLM agent that can read full files, search the codebase, and generate line-by-line comments. It includes a built-in ruleset detecting NPE, thread-safety issues, XSS, and SQL injection — without relying solely on the model. The Delegation Mode, incidentally, is the most interesting feature: instead of OCR reviewing the diff itself, it orchestrates your own coding agent (Claude Code, Codex CLI, etc.) to do the review, using the agent's model without requiring separate credentials.
CI/CD integration (GitHub Actions, GitLab CI, Gerrit) and the Session Viewer — which lets you browse review sessions in a browser, mark comments as "fixed" or "ignored" — put OCR closer to tools like SonarQube than traditional linters. For teams already using coding agents day-to-day, OCR could be the quality layer that's been missing. What distinguishes this project isn't the technology itself, but the fact that it was validated in production by Alibaba before becoming open source — reducing the risk of "theoretical problems that only appear in practice."
2. Google Chrome Modern Web Guidance — The documentation your agent actually needs
GoogleChrome/modern-web-guidance-src received 1,875 stars in a single day — the largest individual gain on this list. The premise is deceptively simple: give coding agents a curated collection of 200+ guides on modern web practices (HTML, CSS, Web APIs) so they use native browser features instead of legacy workarounds.
But the problem this project solves is deep. When you hand a repository to a coding agent, it doesn't just use the code being modified — it copies the 2015 patterns that the model's training absorbed. CSS hacks, imperative JavaScript listeners, unnecessary polyfills that bloat the bundle and degrade performance. Modern Web Guidance solves this with a local agentic RAG system: 1.7 MB of markdown organized in a CLI. The agent runs a scoped search, retrieves the relevant guide (~800 tokens), and applies the best practice — without wasting tokens on irrelevant context.
The Chrome team published a technical deep-dive showing the pipeline: 1,029 searches mined from nightly agent evals (Claude Code, Codex CLI, Gemini CLI) fed the semantic database. Version 1.0 launched officially last week, and adoption is already showing impressive numbers. It's a case study in how a platform team can deliver expertise in structured form to the agent ecosystem. The central message is clear: coding agents need specific documentation, not a generic MDN dump — and Modern Web Guidance is the answer.
3. OpenAI Codex Security — The scanner from the source
openai/codex-security gained 754 stars (13,032 total) and marks a symbolic moment: OpenAI is releasing a security tool for the code of its users, not against them. The Codex Security CLI and SDK scan code, investigate vulnerabilities, and validate fixes — all running on the company's own models.
The structure is modular: cloud for connected GitHub repos, a plugin for local use in ChatGPT Desktop, a CLI for repeatable scans in CI, and an SDK for integrating into third-party tools. The CLI accepts `--auth chatgpt` or API keys, supports bulk scans, and integrates with GitHub Actions and GitLab CI. The discovery pipeline includes threat model export, SQLite-based finding dedup, and verify-fix for checking whether a patch actually resolved the vulnerability.
What makes this project relevant to the open source ecosystem is the open TypeScript SDK. Any security tool can wrap the OpenAI scanner without depending on ChatGPT. If the community adopts it, it could create a standard layer of agent-verified vulnerability validation — a step up from traditional regex-based scanners. OpenAI's approach of opening the SDK while keeping the model proprietary creates an interesting hybrid model: open infrastructure, proprietary intelligence.
4. Daily Stock Analysis — LLM-driven stock analysis, zero cost
ZhuLinsen/daily_stock_analysis with 272 stars (71,494 total) is perhaps the most ambitious project on this list: a complete multi-market stock analysis system powered by LLM, with multi-source market data, real-time news, decision dashboard, and automated notification push. The differentiator? It runs scheduled for free — no API cost, assuming you use local open source models or free tiers.
The architecture connects market data sources (multiple brokers and exchanges), aggregates financial news, feeds an LLM for analysis, and generates trading signals. It supports multi-market (Chinese A-shares, crypto, global equities). The fact that it has nearly 72k stars shows massive demand for accessible quantitative tools — especially in the Asian market, where the project originated. What's notable isn't just the technical ambition, but the accessibility: anyone with a modest computer can run the system, democratizing quantitative analysis that was previously exclusive to financial institutions.
5. Sure — Real personal accounting, not a dashboard
we-promise/sure received 944 stars (14,616 total) and brings an almost revolutionary proposition in the personal finance niche: an app built by and for the community, with a sharp focus on privacy and transparency. Nothing about connecting bank accounts to third-party servers — the idea is that users control their own data.
The project positions itself against the financial data economy that defines modern fintechs. While Plaid, Mint, and competitors collect transactions to monetize insights, Sure proposes that personal accounting be done locally with open tools. Ruby is the language choice, which is a deliberate stance against the TypeScript/Node stack that dominates the productivity tool ecosystem. The community behind the project follows an open contribution model, with no institutional investors — meaning roadmap decisions are made by contributors, not VC executives.
6. PlannOtor — The human-agent coordination workspace
backnotprop/plannotator gained 100 stars (10,379 total) and addresses a problem few open source projects have touched: the interface between humans and coding agents. It's neither a code review tool nor project management — it's a "workspace" where you shape, decide, plan, annotate, review code, and send feedback to agents in one click.
The core metaphor is that, in teams using multiple agents simultaneously, the bottleneck isn't the agent itself but the communication between the human team and the agent. PlannOtor tries to solve this with a structured flow of annotation and feedback, where human reviews are converted into instructions that agents understand. It's an early-stage project, but the direction is clear: the next generation of AI-assisted development tools won't be about code — they'll be about coordination. The team behind it has already identified that 70% of AI-assisted development time is spent on communication, not coding.
7. Uptime Kuma — The monitoring tool every team should have
louislam/uptime-kuma gained 61 stars (93,785 total) and remains, without contest, the de facto standard for self-hosted monitoring. If you have a server, a domain, and want to know when something goes down — without relying on Statuspage, Pingdom, or UptimeRobot — Kuma is the answer. Beautiful interface, alerts via Telegram/Discord/email/Pushover, HTTP/TCP/DNS monitoring. Self-hosted, open source, no tracking.
The slow growth (61 stars) doesn't reflect weakness — it reflects maturity. 93,785 stars and a consolidated integration ecosystem. It's not new, but it deserves the spotlight because every developer should know it and many don't. The beauty of the project lies in its simplicity: it's the kind of software that solves one concrete problem without trying to be everything.
8. Buzz (Block) — Distributed communication in a hive-mind style
block/buzz received 95 stars (37,258 total) and is the most ambiguous project on the list: described as "a hive mind communication platform," it comes from Block (ex-Square) and seems to point toward a synchronized, decentralized communication format. With 37k stars growing, the interest is real, though the exact proposition is still being clarified. The name "Buzz" suggests group communication, and the fact that it comes from Block — a company that has already handled distributed fintechs like Cash App — suggests there may be a practical application behind it.
9. Claude Trading Skills — Technical analysis for traders
tradermonty/claude-trading-skills gained 60 stars (4,074 total) and completes the trio of finance projects on this list. Unlike Daily Stock Analysis (which focuses on general market analysis), Claude Trading Skills are specialized skills for Claude Code aimed at traders: technical analysis, charting, economic calendars, screeners, and trading strategy development.
It's the materialization of a trend: Claude Code skills are verticalizing. Just as skills existed for code review and debugging, now they're emerging for financial analysis. The target audience is clear — traders who already use Claude Code as an assistant and want dedicated skills for market analysis. The project shows how AI tools are specializing by domain, no longer by generic function.
10. Tailcat — Netcat over Tailscale, without the control plane
tailscale/tailcat snagged 324 stars (9,368 total) and solves a niche but frustrating problem: you want to `nc` (netcat) between two hosts on your Tailscale network, but don't want the overhead of the control plane. Tailcat is a simple UDP server that routes packets directly through Tailscale's data plane, without going through the control server. Go, low-level, straight to the point.
For Tailscale users in production, this eliminates latency and a single point of failure. It's the type of tool that 99% of people will never need, but the 1% who do will be grateful it exists. The project shows how the Tailscale community is building layers of utility on top of the base infrastructure.
---
What the overall picture reveals
Three overlapping trends emerge from this edition of GitHub Trending, and together they paint a portrait of the development ecosystem in 2026:
Coding agents are maturing from tools into platforms. Google's Modern Web Guidance and OpenAI's Codex Security are not apps — they're infrastructure. They don't solve one isolated problem; they provide the foundation so that other agents can do better work. It's the difference between selling fish and building the market. When companies like Google and OpenAI invest in tools that empower other agents, they signal that the next wave of innovation comes from the top of the chain, not the bottom.
Personal finance and trading are migrating to open source. Sure, Claude Trading Skills, and Daily Stock Analysis show that the anti-data-tracking movement in personal finance is gaining real momentum. Users want real control over their data, not pretty dashboards powered by third-party servers. The personal finance niche, historically dominated by startups with questionable monetization models, is being reshaped by open source communities focused on privacy.
The next frontier is human-agent coordination. PlannOtor is an early signal. As more people use multiple agents simultaneously, the bottleneck shifts from model capability to the communication interface between human teams and agents. Tools that facilitate this interface will define the next generation of productivity. The question that remains: who will win the race for the best coordination interface — established companies or open source projects born from real user need?
Sources: GitHub - alibaba/open-code-review, Chrome for Developers - Modern Web Guidance, GitHub - GoogleChrome/modern-web-guidance-src, GitHub - openai/codex-security, GitHub - ZhuLinsen/daily_stock_analysis, GitHub - we-promise/sure, GitHub - louislam/uptime-kuma, GitHub - block/buzz, GitHub - tradermonty/claude-trading-skills, GitHub - tailscale/tailcat
✓ Independent sources cross-checked and verified before publishing
