GitHub listed twelve projects this Monday morning, and eight of them were featured here in the past week, so they stay out. Today's four picks are all first appearances: ggml, the tensor library under llama.cpp, which just reached 1.0.0; NVIDIA's SkillEvaluator for testing whether agent skills actually help; Claude-BugHunter, a skill bundle for authorized bug hunting; and Google's Stitch skills for design work with coding agents. Curation order reflects practical interest, not GitHub rank. Trending is not a release date or an endorsement.
Picks are listed in curation order, not GitHub’s rank; each card shows the rank GitHub displayed when the page was checked. Stars measure attention, not security, quality or fitness for use, and no project here was installed or run. “Stars today” is GitHub’s own displayed figure for its daily window, not an audited rolling 24-hour count.
Pick 1 · Machine learning infrastructureGitHub rank #8 of 12
What it does. The tensor library behind llama.cpp: plain C/C++ with no dependencies, it runs machine-learning models on CPUs, GPUs, NPUs and in the browser, with 2- to 8-bit quantization and the GGUF file format.
Caveat. 1.0.0 is a breaking release for anyone on the old API: deprecated functions and the legacy DIAG_MASK ops are gone, and the RPC protocol bump rejects older peers. Core changes are made in the llama.cpp repository and synced here, so issues and pull requests belong there.
Attention, not endorsement
GitHub showed
“43 stars today” as displayed for GitHub’s daily window; not an audited rolling 24-hour count
What it does. A test bench for AI agent skills from NVIDIA: it checks a skill folder for safety and format, flags duplicated guidance, then runs an agent with and without the skill in a sandbox to measure whether it helps.
Caveat. NVIDIA labels it experimental, with community best-effort support and no SLA. Static checks run without a model, but full checks fetch extra security scanners, and live evaluation needs a model provider, a supported agent and Docker; model and sandbox use can cost money.
Attention, not endorsement
GitHub showed
“122 stars today” as displayed for GitHub’s daily window; not an audited rolling 24-hour count
What it does. A skill bundle for authorized bug hunting and external red-team work: 83 skills and 15 slash commands for Claude Code, also loadable by other agents, covering recon, web vulnerability classes, enterprise platform attack chains, validation and report writing.
Caveat. This is offensive tooling, for assets you own or are authorized to test. It is a single maintainer's project with commercial sponsors, its 'battle-tested' claims are unverified, and the copy installer writes into the skill folders of several agents and can wire in a Burp MCP server.
Attention, not endorsement
GitHub showed
“98 stars today” as displayed for GitHub’s daily window; not an audited rolling 24-hour count
What it does. Agent skills for Google Stitch, an AI design tool: they let a coding agent generate and edit screens, extract a DESIGN.md from existing code, and turn Stitch designs into React, React Native or shadcn/ui components.
Caveat. Every skill needs the Stitch MCP server and a Stitch account, so design data goes to Google's service. The README says it is not an officially supported Google product, and the repository's last push was on 17 August.
Attention, not endorsement
GitHub showed
“247 stars today” as displayed for GitHub’s daily window; not an audited rolling 24-hour count
Every repository GitHub listed, in GitHub’s order, with the figures it displayed when the page was checked (): 4 picked today, 8 covered in an earlier digest, 0 not covered. For projects covered before, growth compares the total stars at our first pick with today’s displayed total.
TypeScript · 988 stars today · 26,609 stars in total
Build your own AI office with Univer Office SDK. Bring editing, calculation, collaboration, and file processing into your product for people and AI agents. Built on an extensible open-source core.
Covered before — first picked 15 Nov 2026 at 25,093 stars; now 26,609: +1,516 stars (+6.0%) in 1 day.
TypeScript · 247 stars today · 9,464 stars in total
A library of Agent Skills designed to work with the Stitch MCP server. Each skill follows the Agent Skills open standard, for compatibility with coding agents such as Antigravity, Gemini CLI, Claude Code, Cursor.
TypeScript · 866 stars today · 12,055 stars in total
The workspace to stay in the loop and on top of agents - shape, decide, plan, annotate, review code, share with team, and send all feedback to agents with one click.
Covered before — first picked 14 Nov 2026 at 10,567 stars; now 12,055: +1,488 stars (+14.1%) in 2 days, about 744 a day.
Self-hosted workflows for coding-agent delivery. Orchestrate CodeBuddy, Codex and Claude Code with scripts and human approval gates. Includes Local workflows for coding assistants.
Covered before — first picked 15 Nov 2026 at 1,016 stars; now 1,851: +835 stars (+82.2%) in 1 day.
A Claude Code skill bundle for bug hunting and external red-team work - 82 skills, 15 slash commands, 681 disclosed-report patterns curated across 24 core vulnerability classes, plus enterprise identity + infrastructure attack matrices.
Claude Code skills for equity investors and traders — market analysis, technical charting, economic calendars, screeners, and trading strategy development.
Covered before — first picked 14 Nov 2026 at 4,309 stars; now 5,798: +1,489 stars (+34.6%) in 2 days, about 744 a day.
Multi-tier framework for evaluating AI agent skills with quality gates, semantic overlap detection, synthetic evaluation dataset generation, and live agent evaluation that measures how skills affect agent behavior.
Every project picked in an earlier digest, biggest growth first. “Then” is the total at our first pick’s metadata check; “now” is GitHub’s API star count checked . Growth measures attention, not quality or safety.
GitHub's daily listing held 12 rows at capture. The capture parsed exactly twelve listing rows and is GitHub's own list, not a truncation by us.
All 12 listing entries were inspected. Exact public GitHub API metadata and README files of the four new candidates were read between 07:01:13 and 07:01:19 UTC, and recent releases between 07:01:41 and 07:01:48 UTC. Every checked repository was active and unarchived.
Eight entries featured in the past seven days were left out: GoogleChrome/modern-web-guidance-src (12 November), backnotprop/plannotator, ChromeDevTools/chrome-devtools-mcp, alibaba/open-code-review and tradermonty/claude-trading-skills (14 November), and dream-num/univer, ag-ui-protocol/ag-ui and Tencent/LoopForge (15 November). Their releases since those picks are patch builds, dated routine releases or a release candidate, or none at all, which we did not judge a material change.
Claude-BugHunter is dual-licensed by its README and NOTICE: MIT for code and CC BY 4.0 for skill content and documentation. The license field on its card shows GitHub's detection (MIT) only.
The preceding seven calendar dates, 9-15 November, were checked for local history; all seven editions exist.
The official listing was captured at 08:00:56 Europe/Paris. Literal stars-today labels are GitHub's displayed period figures; API total stars and README evidence were checked separately at the times shown per card. They are not an independently audited rolling 24-hour growth measurement.
Feature descriptions, benchmark, adoption and performance figures come from each project's own README and release notes and are the maintainers' claims. No candidate code, installer, CLI, plugin, skill or model was installed, downloaded or run, and none of this is a security audit or endorsement.