Havuz

This project is for learning: it shows how an agent loop works underneath, in a way more intuitive and visual than reading about it. Behind

🐳 whalegate — OpenAI-compatible gateway into DeepSeek web sessions (Rust): account pool, failover, SSE, native-style tool calling (tool_choi

Think only when the tool call looks unsure: confidence-gated reasoning for small local models. Code, BFCL results and paper.

Fault-injected tool environments with side-effect grading for evaluating and post-training LLM agents

A local AI agent built from scratch in Python with Ollama: conversation memory, system prompts, and multi-step tool calling.

Does the model pick the right tool? Tool-routing tests for your MCP server: run them on Claude and OpenAI models, see where each request lan

Measure and safely cut the token cost of LLM tool and MCP schemas, with equivalence checking and a CI gate.

Local gateway that routes coding-agent tool decisions through Jev and can attach a read-only advisor (GPT or Claude).

Add AI agents to your .NET app: your methods become tools with role checks and human confirmation. Use the ready-made chat widget in Blazor

Review what your agent changed and approve what it may do: React components for multi-file diff review and tool approvals that return a resu

On-device AI agent for Android, tuned for free OpenRouter/Kilo models with automatic model/key fallback and rate-limit recovery. Tools: web

AI inference gateway and control plane for multi-provider routing, policies, caching, tools, and observability.

Catches multilingual voice agents placing the wrong order. τ-bench style benchmark + CI gate for voice ordering agents in English and Hingli

Verified catalog of 300+ APIs and MCP servers with opinions and statistics so that AI agents can call after a one-time API key/OAuth setup.

Open-source AI agent framework for PHP. Tools, memory, security guards and a ReAct loop built in. Works with Laravel, Symfony, WordPress, Dr

Research design for a Jev-native agent system:enable more options than jev provided with virtulization and paging, tool integration, externa

⚡ Millisecond System-1 judgement for every tool call in DeepSeek Harness — Jev-powered risk classification & evidence-gated auto-approval. F

248 small Jev decisions for JavaScript and TypeScript: route, rerank, gate, grade, compare, and label text for agents, RAG, support, code re

Coding-agent tools on TypeSafe's Jev, measured in public. review-router: caught 13 of 13 CVE fixes a filename rule sent to a quick review. j

Wiggle, Claude’ın kod çalıştırma ortamına benzediği iddia edilen araç ve paket setini Docker image olarak yerelde çalıştırmayı sağlar.

Neden yükseliyor · Claude’ın kod çalıştırma ortamının yerelde kullanılabilir bir kopyası olduğu iddiası ve beraberindeki araçlar ile skill dosyaları ilgi çekici. Ancak 23 günlük geçmiş, 91 yıldız, son 28 günde 7 commit ve tek farklı yazar, benimseme ya da sürdürülebilirlik konusunda güçlü kanıt oluşturmuyor.

The execution safety layer for AI agents.

The best model under 200B at near 10 token/s exact batch throughput on a single laptop CPU, with an automatic 8 GB RAM path. Native C, no GP

AI-driven intelligent clipboard turns scattered content across applications into a searchable, editable AI context. It features a clean, ele

轻量的 AI Agent 核心：给它模型地址和工具（文件目录或内存对象），自动循环「问模型 → 执行工具 → 再问模型」；支持 4 种协议、上下文压缩保留、提示词缓存，纯对话模型也能用工具。

Zero-dependency guardrail framework for AI agents. Stops runaway loops, prompt injection, hallucination and unsafe tool calls before they ex

Qwen3.8-27B on one laptop CPU: up to 2.52 token/s, 8 GB tested, no runtime accuracy loss. Native OpenAI-compatible API with function tools;

Progressive tool discovery for DeepSeek Harness — tiny stable surface, searchable catalog, real pipeline execution, context cache intact.

A coding agent harness built specifically for Motif-3

Provenance-aware runtime security for AI agents with deterministic ALLOW / REVIEW / BLOCK enforcement.

Use your Cursor subscription as an OpenAI-compatible API. Direct agent.v1 protocol — real tool_calls, streaming, multi-account. No cursor-ag

Run the native 284B-A13B DeepSeek-V4-Flash-0731 LLM locally on one laptop CPU: pure C, 8 GB RAM minimum, no GPU, best TPOT 0.892 s/token, re

Small local-first agent kernel: bounded tools, reversible edits, auditable runs. CLI + Web, Ollama + compatible APIs.

Lobster0 — a self-hosted OpenClaw-inspired AI agent with persistent memory, governed autonomy, and self-evolution. 受 OpenClaw 启发的自托管 AI Agen

Official RollingGo Hotel Booking MCP Server. 2M+ hotels at wholesale rates, 40+ LLM compatible. Free unlimited calls for businesses & indivi

Macaw — on-device AI agent for macOS. 2.7B LFM2.5 fine-tune, MLX 4-bit, 97 macOS tools, no cloud.

Privacy and security guardrails for Spring AI: request-scoped PII tokenization, least-privilege disclosure to tools, and tool authorization

Hermes Talk: realtime duplex voice for Hermes Agent. Rides your ChatGPT or X Premium subscription, no API key needed. OpenAI, Grok, Gemini.

Self-hosted AI gateway with a built-in dashboard. OpenAI-compatible API, multi-tenant, sandboxed code execution, MCP integrations, file stor

OnDevice Local AI Studio — run LLMs on iPhone offline. SwiftUI workbench with MLX, llama.cpp, whisper.cpp, RAG, and authenticated OpenAI/Ant

MIT-licensed TypeScript runtime for AI-native game characters and worlds, with durable state, reliable actions, memory, tools, and Unity, Go

Open-source local MCP runtime that checks AI-agent tool calls, gates risky actions on approval, verifies results, and avoids blind retries a

Rust AI Agent framework with MCP support, extensible tools, streaming & multi-agent orchestration

Unofficial OpenAI-compatible bridge for Factory Droid

An open-source, self-hosted social AI agent across QQ, Telegram, Feishu/Lark and personal clients, with persistent identity, long-term memor

Janux - One component, two faces: UI for humans, MCP tools for AI agents. The agent-native fullstack framework.

Losslessly compress the tool output your AI agent sends the model, prove every value survived, and report the exact tokens removed. Run it a

The vision plugin for OpenCode that truly understands images. Inspect, read, and reason about any screenshot or picture with deeper understa

MCP gateway with GraphQL tool aggregation

Provider-neutral bounded agent runtime for Android/JVM: controlled context (trust/priority/budgets), profile-scoped tools, OpenAI-compatible

Mindlas catches your coding agent drifting before the bad code lands. Real-time tracking and correction for context rot, unverified done cla