编辑甄选 · OSS

GitHub 高星项目· 4 projects

由潜龙编辑团队甄选，每周更新最值得看的开源项目。我们看重的不是 Star 多少，而是它解决问题的方式。

全部主题高潜项目 agent-skills agents ai-agents ai-tools anthropic chatgpt claude claude-code claude-skills cursor developer-tools gemini ai ai-gateway api-key-management budgets cost-tracking gateway litellm-alternative llm llm-gateway llm-proxy llm-routing agent-security agentic-security audit-log causal-proof-engine llm-security local-first mcp-security agentic-ai ai-agent ai-coding ai-coding-assistant autonomous-agents cli code-assistant coding-agent agents-md codex llm-tools mcp pm-tools product-management acp-client acp-gateway agentic-workflow ai-memory claude-plugin dsh-plugin gemini-cli-extension hermes-agent digital-twin knowledge-graph memory multi-agent personal-ai rag agent-teams ai-automation autonomous-agent aws business-operations claude-code-plugin devops ecommerce agent-infrastructure agent-platform agent-runtime agents-api ai-infrastructure cloud-agents durable-agents durable-execution managed-agents ai-assistant langchain langchain-python local-llm ollama privacy react-agent sovereign-ai android computer-use cross-device desktop-app gui-agent ios linux meta-muse ai-methods api docker fastapi mthds pipelex a2a agent-evaluation llm-agents llm-evaluation model-context-protocol python reinforcement-learning reinforcement-learning-environments sandbox agent code-generation code-intelligence compiler cross-platform harness harness-engineering no-code openai orchestration xinference agent-swarm automation bun code-review github guardrails ai-workstation comfyui local-ai mcp-server ai-pentesting attack-surface-management bug-bounty cybersecurity ethical-hacking ethical-hacking-tools exploitation kali-linux agent-orchestration agentic-workflows ai-software-engineering go react self-hosted software-factory workflow-automation corecoder deepseek apple-silicon inference-engine llm-inference macos metal mixture-of-experts mlx moe framework google-antigravity graphrag knowledge-management cheating cluely cluely-alternative cluely-clone final-round-ai agent-framework artificial-intelligence open-source rust hacs home-assistant home-automation ai-hedge-fund ai-quant byok docker-compose financial-ai fintech investment-research llm-agent c coding-assistant llama-cpp openrouter repl agentic-coding ai-governance context-engineering context-gateway context-intelligence copilot csharp cuda cuda-programming dotnet dspark gemma gemma4 gguf glm inference analytics antigravity cli-tool cursor-ai insights kiro large-language-models ai-workbench data-analysis deepseek-harness dsh electron batch-processing content-extraction data-collection firecrawl firecrawl-ai javascript-rendering agent-client-protocol codex-cli course-downloader download-manager downloader hotmart-downloader instagram-downloader ai-note-taking ai-second-brain claude-code-skill claude-memory claude-skill karpathy-llm-wiki ai-orchestration business-automation enterprise-ai erp global i18n pi pi-agent pi-desktop plugins ai-platform chatbot conversational-ai generative-ai llmops operating-system claude-code-skills multi-model runtime ai-identity digital-soul gemini-cli long-term-memory agent-memory copy-on-write embedded-database full-text-search hnsw hybrid-search llamaindex mysql oceanbase ai-sdk human-in-the-loop a2a-protocol agent-cli ai-agents-cli ai-workforce llm-runtime local-ai-agents agent-harness agent-observability agent-plugin agent-sandbox audit-logging amd gfx1151 gpu-inference hip llm-serving long-context openai-api qwen acryl agent-context-relay agentic agentic-development-environment ai-agent-permissions ai-safety api-gateway audit-trail authorization decionis execution-authority agent-workspace apple-intelligence cpp diffusion-models edge flutter kotlin llamacpp multimodal electron-app gui embeddings knowledge-base semantic-search agent-os grok-bot-alternative muse-alternative agent-loops data-catalog data-governance data-integration data-lineage data-pipeline data-platform database-management er-diagram internal-tools database-client database-gui database-tool mariadb postgresql ai-animation ai-video-generator capcut claude-agent-sdk content-creation ffmpeg image-to-video multi-provider novel-to-video markdown muapi bm25 coding-agents git-native golang llm-memory ai-code-review devtools

共 4 个项目·第 1 / 共 1

排序Star 最多最新发布

ArcReel

AI Agent 驱动的视频生成工作台

4.9k

这个方向最近挺热，ArcReel 试图解决的是 AI 视频生成中“工作流断裂”的问题。它不是一个单一的视频生成模型，而是一个可以私有化部署的 AI 视频工作台。简单说，它想把小说或剧本直接转化为角色、场景、分镜，最终输出视频甚至剪映草稿。这里有个小看点，它特别强调了跨镜头的一致性以及多模型供应商的...

Python高潜项目971

Pipelex

AI 方法构建与部署工具

939

简单说，它想把 AI 能力的封装和部署变得更标准化。你可以用编码代理构建特定的“AI 方法”，然后将它们打包运行在任何地方——无论是作为聊天机器人的 MCP 服务器、面向用户的 Web 应用，还是供其他软件调用的 API。这个方向最近挺热，原因也不难理解：开发者需要一种更干净的方式来管理 AI 逻...

Python高潜项目79

本

sandbaseai

本地优先的智能体沙盒运行时

648

这个项目定位为一个本地优先、可自托管的 AI 智能体运行环境，同时也是一个 MCP（模型上下文协议）桥接器。简单说，它想为 AI 智能体提供一个安全的执行底座，包含了沙盒会话、记忆管理、凭证存储以及审计和重放功能。真正难的地方不是让 AI 跑起来，而是如何控制它在本地环境中的执行边界。它提供了一个...

TypeScript高潜项目69

decionis

agent-safe-pipeline：AI 智能体执行权限网关

533

当 AI 智能体开始拥有执行真实世界操作的能力时，安全和权限控制就成了核心痛点。agent-safe-pipeline 提出了一个执行权限网关的思路，专门用来拦截和审查智能体发出的关键 HTTP 请求。它不像是一个普通的 API 路由，而更像是一个带有审计追踪的“海关”。当智能体试图执行敏感操作时...

TypeScript高潜项目57