Search

AI model: aimeigaoshou/agent-jev

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

A small Kotlin/JVM workshop project for learning how to build: MCP servers over stdio an MCP client that launches and connects to a local server process an agent loop that lets Gemini discover MCP tools and call them dynamically simple data integrations against football and cryptocurrency APIs The repository is structured as an educational sample

手机上的 AI Agent：自然语言驱动 103 个内置工具直接操作本机——文件、应用、抓包、APK 反编译、Linux 环境、界面自动化。内置 MCP 客户端与 MCP Server。

Put a personal agent inside your product, at 1/1000 the cost. Personal agent infra that runs VM-less personal agents on Cloudflare: Agent Mail, WeChat, Telegram, cloud browser, long-term memory.

Provenance-retaining cross-session context merge for DeepSeek Harness — a confluence theorem with a counterexample, and a benchmark for it.

Context compaction for RAG and agents: relevance-score every chunk with a local System 1 decision model and cut tokens before they reach the LLM

Add AI agents to your PHP app in an afternoon - local-first, no Python sidecar, no framework to learn.

MCP-сервер для hh.ru: поиск, резюме, отклики, переписка с работодателями и скрытая аналитика по одному аккаунту соискателя через собственный JSON сайта (без официального API). Исходники открыты, некоммерческая лицензия.

Evals as code for LLM apps and agents. Write tasks and graders in YAML, run them in CI, fail the build on regressions.

A local-first, multi-agent writing assistant with project knowledge tools

Reusable LLM security scanning library — language-agnostic guardrails for AI agents

A general immune system for LLM applications: two lines, every call, reflexes at every boundary.

Run headless pi agent sessions in-process with local socket clients, durable replay, live streaming, and explicit driver arbitration.

A suite of test scenarios for multi-agent reinforcement learning.

Measure whether your coding agent actually follows the rules in AGENTS.md / CLAUDE.md.

Automated research loop that drives WorldQuant BRAIN alphas to all 8 submission gates. Generate, guard, simulate, evaluate, refine — never submits.

Tool search for LLM agents, ranked by System-1 decision models, with a benchmark that measures prompt caching.

Local, private token and cost atlas for AI coding agents (Claude Code, Codex, OpenCode, Pi)

AgentSchool CLI —— 让 AI Agent 一行命令接入 AgentSchool：抖音解析 / 语音转写 / 文件平台 / 公众号创作（as / agentschool 双入口，Agent Skill 公开仓分发）