Agent Skills
stareezy-1/frontend-architecture-skill
stareezy-1/frontend-architecture-skillSource for the frontend-lighthouse skill - portable Lighthouse CI Core Web Vitals gates, performance budgets, and GitHub Actions reporting (MIT)
1 of 50
1 of 50
Assistants & Domain Agents
forge
antoinezambelli/forgeA Python framework for self-hosted LLM tool-calling and multi-step agentic workflows
2 of 50
Frameworks & SDKs
ruby_llm
crmne/ruby_llmThe Ruby AI framework.
3 of 50
Harnesses & Runtime Infra
Respan Gateway
respanai/respan-gatewayOne endpoint for every LLM vendor. Any SDK, any model, any tool. One Rust binary
4 of 50
Plugins, Themes & Clients
Turbo-IO
Turbo1123/Turbo-IO雷鸟 iO / RayNeo iO 非官方 SDK：原生鸿蒙 HarmonyOS SDK/客户端、iOS 客户端及 iOS/Android 官方 App 扩展。自有模型、联网搜索、眼镜导航、录音研究、Codex 与 Web 预览，各平台能力以文档为准。面向开发者，仅供非商业学习研究；只提供源码，不提供 IPA/APK/HAP 或签名包。
5 of 50
Orchestrators & Multi-Agent
statewright
statewright/statewrightState machine guardrails that constrain which tools an agent can call in each phase of a workflow, turning open-ended loops into deterministic state transitions. The research result is striking: local models went from 2/10 to 10/10 passing on a SWE-bench…
6 of 50
Coding Agents
Open Session
tellahq/opensessionSelf-hosted server driving coding sessions in git worktrees on your own box or in isolated sandboxes, with a web UI, Slack/Linear/Plain/GitHub intake, diff and PR review, and multiple Codex and Claude subscriptions.
7 of 50
Sandbox, Security & Governance
agent-sentinel
junlinwk/agent-sentineleBPF/BPF-LSM prototype that monitors local agent behavior and atomically blocks prompt-injection-driven access to sensitive files at the kernel boundary
8 of 50
Observability & Evals
claude-code-status-bar
briansmith80/claude-code-status-barby briansmith80 - Configurable status bar for Claude Code: usage limits with pacing markers, context window, git state, live activity, session cost, and 8 colour themes. Pure bash, zero dependencies
9 of 50
Research & Data Agents
OpenRaiser/NanoResearch
OpenRaiser/NanoResearchEnd-to-end autonomous research engine that plans experiments, generates code, runs jobs locally or on SLURM, analyzes real results, and writes papers grounded in those outputs
10 of 50
Creative, Voice & Media
Enclave
yuanzui0728/EnclaveSelf-hosted single-owner AI social world; each instance is populated by autonomous AI residents with personalities, schedules and relationships who chat, form group conversations, post to a social feed and proactively message the owner
11 of 50
Docs, Learning & Lists
awesome-agent-architecture
hardness1020/learn-agent-architectureTrilingual, section-by-section architecture notes on modern agent harnesses: loop engineering, tools, permissions, context, memory, multi-agent coordination, and evaluation, studied through real systems such as Claude Code and Hermes Agent, with runnable…
12 of 50
Context, Memory & RAG
funes
huggingface/funesby David Corvoysier - Durable cross-agent memory for Claude Code, Codex, pi, and Hermes. Your memory is a Hugging Face dataset you own: recall it from any machine or share it with your team. Single binary, local index, pinned local embedding models
13 of 50
MCP Servers
damientilman/mailchimp-mcp-server
damientilman/mailchimp-mcp-serverMailchimp Marketing API: manage campaigns, audiences, reports, automations, landing pages, e-commerce data and batch operations
14 of 50
Agent Skills
cinetic
Leonxlnx/cineticCinematic launch films and motion design from code, directed by your coding agent.
15 of 50
Assistants & Domain Agents
Autobot
crystal-autobot/autobotCrystal - AI assistant with kernel-enforced sandboxing, multi-provider LLM, voice, vision, cron, and MCP tools
16 of 50
Frameworks & SDKs
Stackwise
stackwiseai/stackwiseThe open source AI app collection
17 of 50
Harnesses & Runtime Infra
Mosec
mosecorg/mosecA machine learning model serving framework with dynamic batching and pipelined stages, provides an easy-to-use Python interface
18 of 50
Plugins, Themes & Clients
EchoBird
edison7009/EchoBird(3,416 ⭐) - Desktop app for switching ChatGPT, Codex CLI, and Claude Code accounts, tracking quotas, and routing model requests with automatic failover
19 of 50
Orchestrators & Multi-Agent
claude-consensus
tonydzi/claw-consensusConsensus protocol for LLM agents running on separate machines: propose/counter/accept/commit rounds, a dual-rail message bus with ACK tracking, and self-healing sync to keep agent state from drifting
20 of 50
Coding Agents
repomon
AliHamzaAzam/repomonRust TUI that supervises a fleet across many repositories at once, in durable tmux sessions you can approve from your phone.
21 of 50
Sandbox, Security & Governance
SkillTotal
pezhik/skilltotalStatic, offline scanner for AI components (MCP servers, agent skills, npm/PyPI packages, repos): supply-chain risk, dangerous capabilities, prompt-injection, exfiltration; deterministic, evidence-anchored, SARIF + pre-commit/GitHub Action
22 of 50
Observability & Evals
Terminal-Bench
harbor-framework/terminal-bench-1The de facto benchmark and execution harness for terminal agents: real sandboxed shell tasks verified by deterministic test scripts, now at 2.x on the Harbor framework. Cited throughout this list because it is the yardstick every harness-only improvement…
23 of 50
Research & Data Agents
eimenhmdt/autoresearcher
eimenhmdt/autoresearcherEarly open-source package for automating scientific workflows, currently centered on literature-review generation with an ambition toward broader autonomous research
24 of 50
Creative, Voice & Media
SkyAGI
litanlitudan/skyagiEmerging human-behavior simulation capability in LLM agents. github
25 of 50
Docs, Learning & Lists
Awesome AI Agents
alternbits/awesome-ai-agentsAI agents for automation and development
26 of 50
Context, Memory & RAG
agent-memory-vault
mcncarl/agent-memory-vaultMarkdown-first shared memory vault for Claude Code and Codex with SQLite, Zvec, Git, closeout, and audit
27 of 50
MCP Servers
devag7/linkedin-mcp
devag7/linkedin-mcpLinkedIn via an authenticated browser session: profiles, people/job/company search, feed, messaging and gated writes (connect, post, react, comment)
28 of 50
Agent Skills
Context Engineering Kit
NeoLabHQ/context-engineering-kitSpec-driven development workflow that transforms prompts into production-ready implementations through structured planning, architecture design, and LLM-as-a-Judge based quality gates
29 of 50
Assistants & Domain Agents
Self-operating computer
OthersideAI/self-operating-computerSOC is a framework enabling multimodal models to operate a computer using human-like inputs and outputs, with compatibility for various models such as GPT-4v, Gemini Pro Vision, and LLaVA, offering future support for additional models and featuring various…
30 of 50
Frameworks & SDKs
Speechless
uukuguy/speechlessLLM based agents with proactive interactions, long-term memory, external tool integration, and local deployment capabilities
31 of 50
Harnesses & Runtime Infra
kimi-cc
LLM-Red-Team/kimi-cc(1.7k ⭐) - Use Kimi's latest model (kimi-k2-0711-preview) to drive Claude Code
32 of 50
Plugins, Themes & Clients
agent-sdk-dev
ccplugins/awesome-claude-code-pluginsphoto-led short-form vertical reels (Instagram/TikTok/YouTube Shorts/Facebook) via MCP, $0.06/render, 50 free seeds, no subscription
33 of 50
Orchestrators & Multi-Agent
toryo
JesseRWeigel/toryoTrust-based delegation with quality ratcheting that commits improvements and reverts regressions. Chains Claude Code, Aider, Gemini CLI, Ollama.
34 of 50
Coding Agents
ripwire
redhat-et/ripwireThe ripgrep of AI context: a zero-dependency C++23 CLI + MCP server for coding agents. Find what you want without reading the repo, then check you built what you meant — blast radius, tests-to-run, quality deltas. Signatures at 74.7% fewer bytes than bodies…
35 of 50
Sandbox, Security & Governance
Llm_Agents_Security
johnny22245/llm_agents_securityThis LLM agents deals with establishing AI security and safety
36 of 50
Observability & Evals
Ai-Agent-Design-For-Iq-Tests
tex216/AI-Agent-Design-for-IQ-TestsDesigned an AI agent to pass Raven’s Progressive Matrices tests, details see the Final-Project-Report
37 of 50
MCP Servers
kehvinbehvin/json-mcp-filter
kehvinbehvin/json-mcp-filterQuery and extract only the needed parts of JSON files to keep LLM context small
38 of 50
Agent Skills
sno-station-skills
sno-ai/sno-station-skillsSkills for Claude Code and Codex: cross-vendor peer review, handoff when one agent hits its usage limit, delegation and reporting
39 of 50
Assistants & Domain Agents
Git AI
git-ai-project/git-aiGit extension that tracks AI-generated code and the prompts behind each line
40 of 50
Frameworks & SDKs
agentlego
InternLM/agentlegoEnhance LLM agents with versatile tool APIs
41 of 50
Harnesses & Runtime Infra
ZenML
zenml-io/zenmlOpen-source framework for orchestrating, experimenting and deploying production-grade ML solutions, with built-in langchain & llama_index integrations
42 of 50
Plugins, Themes & Clients
8B AI Website Builder
8bsite/8b-agent-pluginGemini CLI extension that builds an animated one-page website: the agent picks an 8B generated design, writes the copy and returns a preview link, plain-word edits and one HTML file. Remote MCP https://mcp.8b.com/mcp, no account or API key. Install: gemini…
43 of 50
MCP Servers
ariekogan/ateam-mcp
ariekogan/ateam-mcpBuild, validate and deploy multi-agent AI solutions on the ADAS platform: design skills with tools and manage the solution lifecycle
44 of 50
Agent Skills
Necmttn/ax
Necmttn/axSource for the ax-extract-workflow skill - reconstruct workflow behind past coding-agent artifacts using local ax sessions, commits, skills, and tool traces (AGPL-3.0-only)
45 of 50
MCP Servers
duriantaco/skylos
duriantaco/skylosDead code detection, security scanning and code quality analysis for Python, TypeScript and Go, with AI-assisted remediation
46 of 50
MCP Servers
tomyangod/architecture_viewer
tomyangod/architecture_viewerLocal architecture gate and CLI that diffs AI edits against git HEAD to detect new cross-layer dependencies and structural changes
47 of 50
MCP Servers
AirSaiga/Precis
AirSaiga/PrecisLocal-first data quality workbench for Excel/CSV: validate against schemas and 10 zero-code constraint types (uniqueness, ranges, foreign keys, allowed values and more), with row-level error reports. Also ships a visual desktop app, CLI and REST API
48 of 50
MCP Servers
aos-standard/mcp-blast-radius
aos-standard/mcp-blast-radiusStatic blast-radius extraction and an AOS manifest.json divergence check (Oracle, Permitted, Prohibited), also usable as a blocking CI gate
49 of 50
MCP Servers
kucherenko/jscpd
kucherenko/jscpdDuplicate code detection with jscpd: check snippets and files for copy/paste duplication against a scanned project before committing
50 of 50
How these 50 were chosen
Every category gets slots in every day, weighted by the square root of its size — so the 14 categories here are all represented, the largest one does not take half the page, and a category with a few dozen projects in it is not crowded out by one with thousands. Within a category, the projects that have gone longest without appearing go first. Stars are not an input: the front page already ranks by them, and a Discover that consulted them would be the front page with a different heading.
The week's seven days are picked at once, on Sunday, and written to discover.json — so today's 50 are the same 50 for everyone, they can be linked to, and they are still here when you come back this afternoon. The day turns over at midnight, Chicago time.