cloudflare / capnweb
JavaScript/TypeScript-native, low-boilerplate, object-capability RPC system
cloudflare / capnweb
Bitterbot-AI / bitterbot-desktop
Bitterbot - a mesh of agents that turns shared experience into collective capability.
opencoredev / social-sdk
Typed TypeScript SDK for direct and managed social platform integrations, with capability checks, explicit outcomes, and deterministic testing.
PKUfudawei / dsh-capability-menu
管理 DeepSeek Harness 中工具与技能的暴露策略，支持按需检索与调用。Manage tool and skill exposure in DeepSeek Harness with on-demand discovery and invocation.
flankerhqd / cyvisguard
Security control plane for AI agents — identity and delegation, capability policy, data-flow taint and a live audit trail, enforced over MCP. Guards a real Claude Code end to end.
OpenLlamas / model-xray
Model authenticity / dilution detection CLI: one command, one report. Detect fake or watered-down LLM endpoints via identity, context, tokenizer, capability and latency checks.
awoaCrim / pi-openai-toolkit
OpenAI capability toolkit for Pi with Codex remote context, Responses compaction, hosted web search, image generation, and tool call review.
UNLINEARITY / dsh-code
A terminal coding interface that brings the DeepSeek Harness kernel's full capability set — special modes, plugins, model & subagent management — to the terminal. Interaction layer inspired by Claude Code, Codex CLI, and Pi agent; model-switch animations; unified panel key grammar. Tracks the latest upstream RC. Self-iterating since 1.0.0.
Prohao42 / deepseek-harness-master
炸炸酥网安 · DeepSeek Harness：一个把渗透测试装进 AI Agent 的开源框架，它是什么 DeepSeek Harness 是一个 插件化的 AI Agent 运行框架 ——在它这里，"一切皆插件"。底层基于 Cordis，上层把会话、工具、LLM、沙箱、技能全部拆成独立 capability，像搭积木一样组合。本项目在此基础上预装了 103 个渗透测试技能 ，让它从通用 Agent 变成开箱即用的网络安全 AI 工作台。
rbayuokt / expo-adaptive-glass
Adaptive native glass surfaces for Expo and React Native. Scales blur, refraction and highlights to device capability and live frame performance.
AtlasCloudAI / awesome-gpt-image-2.5-prompts
Prompts for GPT Image 2.5, written against the capability areas in OpenAI's announcement — each shipped with the image it actually produced.