NVIDIA’s SkillSpector scans agent skills before installation

#1 🐙 NVIDIA/SkillSpector

NVIDIA/SkillSpector - NVIDIA/SkillSpector scans AI agent skills before installation for prompt injection, data exfiltration, supply-chain risks, and other threats using 71 patterns across 17 categories. The Apache-2.0 repository includes source code, tests, documentation, extensions, Docker support, optional LLM analysis, and multiple report formats. Worth trying if you need to assess third-party agent skills before adoption.

#2 🐙 JetBrains/thinkrail

JetBrains/thinkrail - **JetBrains/thinkrail** is an Apache-2.0 agentic IDE built around the pi coding agent, with isolated git-worktree workspaces and concurrent agent sessions. It includes desktop and browser-based CLI apps, an IDE workbench, living specs, reusable skills, workflows, and MCP support; the CLI requires Git and a model provider. Worth trying if you need parallel, agent-assisted development with durable project context.

#3 🐙 heygen-com/hyperframes

heygen-com/hyperframes - heygen-com/hyperframes is an open-source framework for turning HTML, CSS, media, and seekable animations into deterministic MP4 videos, with 7 published skills and about 450 ready-made blocks and components. The heygen-com GitHub account released HyperFrames v0.8.202 on October 26, 2026, adding smoother preview audio fades, whole-number Lambda render scaling, and video-function warming. Worth trying if you need agent-oriented, HTML-based video creation. Release notes

#4 🐙 langfuse/langfuse

langfuse/langfuse - **langfuse/langfuse** is an open-source platform for tracing, evaluating, and debugging LLM apps, with prompt management, datasets, a playground, APIs, an AI gateway, and self-hosting configurations. Langfuse released v4.70.1 on October 26, 2026, updating Node to 24.22.1 and improving mobile topics, media URL renewal, and gateway telemetry shutdown. Worth trying if you need one platform for LLM observability and evaluation. Release notes

#5 🐙 akitaonrails/ai-memory

akitaonrails/ai-memory - akitaonrails/ai-memory is an MIT-licensed, self-hosted app with a server, lifecycle hooks, and integrations that give more than twenty AI coding-agent harnesses shared long-term memory and handoffs; Markdown in a git-backed wiki is the source of truth, and the default path uses zero LLM calls. akitaonrails released v3.2.5 on 2026-10-25 at 18:28:12 UTC; discovery metadata showed 13,641 stars and a 3,254 weekly trending gain. Worth trying if you need memory across agents, projects, machines, or teams. Release notes

#6 🐙 multica-ai/andrej-karpathy-skills

multica-ai/andrej-karpathy-skills - multica-ai/andrej-karpathy-skills packages four Andrej Karpathy-inspired coding-agent principles in a CLAUDE.md file, Claude Code plugin, and Cursor rule: think first, keep implementations simple, make surgical edits, and verify goals. The snapshot reports 224116 stars, a 457 daily trending gain, and a 2937 weekly trending gain. Worth trying if you need reusable guidance for more cautious, focused coding agents.

#7 🐙 n8n-io/n8n

n8n-io/n8n - n8n-io/n8n is a fair-code platform for building AI agents and workflows with a visual canvas, JavaScript, Python, npm packages, self-hosting or cloud deployment, and 1500+ integrations. n8n-io released n8n@2.44.7 on October 26, 2026, with 5 bug-fix entries covering sub-workflow errors, IMAP reconnects, SharePoint mapping, and two task-runner certifi updates. Worth trying if you need flexible AI workflow automation with model choice, human approvals, and observability. Release notes

#8 𝕏

Logan Kilpatrick recapped that managed agents were shipped at Google I/O and described an admin dashboard built on top of them that he said should be useful.

#9 🐙 AgriciDaniel/claude-seo

AgriciDaniel/claude-seo - AgriciDaniel/claude-seo is an MIT-licensed, local SEO plugin for Claude Code covering technical and content audits, schema, AI-search optimization, and agent-readiness checks. The repository contains 26 reusable sub-skills, 19 specialist agents, 9 optional extensions, plus documentation, scripts, schemas, and tests. Worth trying if you need a broad SEO auditing toolkit inside Claude Code.

#10 🐙 alchaincyf/huashu-design

alchaincyf/huashu-design - **alchaincyf/huashu-design** demonstrates Huashu Design 3.0, a reusable agent skill for producing clickable web/app prototypes, editable presentations, animations, and printable infographics. The MIT-licensed repo includes 84 HTML-native styles, scripts, assets, references, demos, and a workflow that develops three of six thumbnail candidates while retaining three alternatives. Worth trying if you need multi-format design exploration inside supported coding agents.

#11 🐙 latent-spaces/brag

latent-spaces/brag - latent-spaces/brag is an MIT-licensed agent skill that turns a project into a short launch video with music, motion, and share copy. latent-spaces released version 0.5.0 on 2026-10-23 at 21:18:23 UTC; the 18,703-star repository includes reusable skills, examples, and Claude and Codex plugin files. Worth trying if you need launch-video automation and can provide Node.js 22+, FFmpeg, and the Hyperframes CLI. Release notes

#12 ▶️

How to Make Amazing Videos with Claude and HyperFrames | Jake Moran

Peter Yang

Also covered by: @Peter Yang

#13 🐙 stablyai/orca

stablyai/orca - stablyai/orca is stablyai’s MIT-licensed orchestrator for running CLI coding agents side by side in isolated Git worktrees; its repository contains desktop, mobile and cloud code, reusable skills, docs and examples. stablyai released Orca v1.4.241 on 2026-10-26 at 03:48:28Z, adding per-agent chat permissions, browser tools, visible agent browsing, SSH recovery and terminal/mobile fixes; it has 106,458 GitHub stars and 9,622 weekly trending gains. Worth trying if you need one workspace for multiple coding agents. Release notes

#14 🐙 pingdotgg/t3code

pingdotgg/t3code - **pingdotgg/t3code** is an MIT-licensed agent control surface shared by pingdotgg, with iOS, Android, web, and Electron desktop apps plus in-repo documentation. It controls six providers—Codex, Claude, Cursor, Grok Build, OpenCode, and Google Antigravity—and has 30,311 GitHub stars and 1,770 weekly trending gains; setup and authentication are required except for Antigravity’s no-CLI flow. Consider it if you need one interface for supported local agents.

#15 🐙 farion1231/cc-switch

farion1231/cc-switch - **farion1231/cc-switch** is an MIT-licensed cross-platform desktop app for managing providers, MCP, Skills, prompts, and sessions across AI coding assistants; its repository includes app source, documentation, and tests and has 148842 stars. farion1231 released v4.0.20 on October 24, 2026, adding Oh My Pi as the 13th managed application, browser login for managed ChatGPT accounts, app grouping and reordering, plus fixes. Worth trying if you need one desktop interface to configure multiple supported assistants. Release notes

#16 🐙 mvschwarz/openrig

mvschwarz/openrig - mvschwarz/openrig is an Apache-2.0 system for defining persistent Claude Code and Codex agent teams, with reusable skills, docs, a demo, Docker assets, and tests. mvschwarz released OpenRig 0.7.2 on October 25, 2026, adding per-seat logins and runtime switching; consider it if you need coding-agent teams with shared context and individually configured seats. Release notes

#17 🐙 nightly-labs/openbot

nightly-labs/openbot - nightly-labs/openbot is a local-first desktop app for persistent Codex, Claude, Grok, OpenCode, Gemini, Cursor, and Cline agents, with workspaces, queues, file transfers, browser control, and agent-to-agent messaging. nightly-labs released OpenBot v0.50.0 on 2026-10-25T14:13:57Z; changes by NorbertBodziony include Ubuntu 26.04 remote-desktop startup support and multiple Google accounts for Gemini. Consider it if you accept development-preview agents running with danger-full-access and no per-action confirmations after first-launch consent. Release notes

#18 🐙 ruvnet/ruflo

ruvnet/ruflo - **ruvnet/ruflo** is an MIT-licensed toolkit for coordinating AI coding agents in Claude Code and Codex, with CLI/MCP tooling, memory, federation, security controls, plugins, documentation, and a terminal console. ruvnet released Version 3.63.1, dated October 25, 2026 and published October 26, as a patch for @claude-flow/cli, claude-flow, and ruflo; no leaf package was published. Worth trying if you need a reusable foundation for multi-agent coding workflows. Release notes

#19 🐙 comet-ml/opik

comet-ml/opik - **comet-ml/opik** is an Apache-2.0 LLM observability and evaluation platform with a self-hostable backend, web app, SDKs, tracing, datasets, experiments, dashboards, guardrails, and agent optimization. comet-ml released Opik 2.2.120 on October 26, 2026, adding thread environment filters, online-evaluation starter cards, and a custom annotator view. Worth trying if you need self-hosted tracing, evaluation, and monitoring for LLM apps or agents. Release notes

#20 🐙 infiniflow/ragflow

infiniflow/ragflow - infiniflow/ragflow is an Apache-2.0 open-source RAG engine combining document ingestion, multi-step retrieval, evidence verification, citations, and structured knowledge generation for LLM applications. infiniflow released RAGFlow 1.0.0-rc1 on 2026-09-29; the 92,891-star repository includes Docker and source deployment files plus pre-built agent templates. Worth trying if you need a self-hosted context layer with configurable retrieval depth.