Python agents

Automate AI agent behavior tuning with human feedback, test small mutations, and keep what improves performance over time

基于 FastAPI、AgentScope 2.x、React、Tavily 与 OpenTelemetry 的多智能体深度研究平台

Sort your documents into a corpus you can query — a filing skill for Claude Code

Official implementation of LifeMem: longitudinal memory for more faithful, diverse, and dynamically evolving LLM agents.

Block irreversible AI-agent actions before they run — deterministic, fail-closed action veto for Claude Code, Codex, Cursor, and other tool-using agents.

Eureka — task-conditioned Meta-Agent orchestration for scientific discovery, developed by ManXis.

Self-hosted audio API in one Docker container. Stem separation, mastering, MIR (BPM/key/chords), audio→MIDI, text-to-music + SFX (MusicGen/Stable Audio/Riffusion/AudioLDM2), diarization, speech enhancement, restoration, embeddings + zero-shot tagging, workflows, async + webhooks. REST + MCP. CPU + CUDA.

Reference implementation of durable execution patterns (signals, queries, saga pattern, idempotency, replay-safe determinism) using Temporal, applied to an agentic triage workflow.

Jango AI , Multi-agent AI system with mobile app, web, and bot interfaces — Tavily-powered search orchestrated across a super-agent + micro-agent swarm.

A high-performance AI coding agent.

Whose Side Is Your Agent On? PrincipalBench — a multi-turn benchmark and post-training methods (prompt scaffold + per-token-KL distillation) for multi-party principal loyalty in LLM agents.

The layer a sandbox doesn't give you: a PreToolUse hook that blocks unapproved spending, account creation and fund movement before the call runs, queues them for human review, and proves it actually ran -- including whether your harness fired it for subagent tool calls. Plus a spend ceiling and loop detector.

Local-first, LLM-agnostic memory system — hybrid retrieval (vector + lexical + graph), multi-signal ranking. Runs on 4GB RAM, CPU-only, no cloud.

AI-Psychiatry is a Claude plugin that gives AI coding agents an executive-function layer to detect and control attention drift, overthinking, recursive reasoning, compulsive verification, scope drift, retry loops, deadlocks, livelocks, context waste, and failure to stop after completion.

Privacy-first AI assistant for legal documents. Ask questions in natural language, get instant answers with exact sources. Scans contracts for risky clauses automatically. Runs 100% on your computer—no cloud required. Supports local AI (Ollama) or cloud APIs (ChatGPT, Claude, etc.). completely Free

Reliability and safety harness for LLM cloud agents. Tool-selection accuracy vs. tool count, blast-radius guardrails, and prompt-injection red-teaming against a real AWS mock. Every number is a committed JSON artifact.

基于 DeepSeek 的模型自驱多 Agent 深度研究系统:自研 harness、四角色动态编排、GLM 独立裁判评估 + 基线对比、Streamlit 实时编排可视化

Linux kernel dev plugin: L1-L4 verification pyramid (checkpatch/sparse/build/boot) + graph-level governance (Goodhart/blind-spot/conflict) + x-kernel feature audit (P1-P12)

Rust dev plugin: fmt-clippy-build-miri/audit verification pyramid + 12 semantic practice cards + problem-oriented docs map

🐉 HYDRA — 7-headed AI bug bounty system for opencode: hunter, verifier, reporter, plan, auditor, recon, debug · 928 skill modules · one-command setup

Multi-agent stock research app built with CrewAI that analyzes news, financials, and SEC filings to generate investment recommendations via web UI/API.

🤖 Multi-agent AI code reviewer & documentation generator powered by CrewAI, Google Gemini 2.5 Flash, and Streamlit.

An agentic AI project for patent discovery, trend analysis, and innovation forecasting using OpenRouter, OpenSearch, and CrewAI.

Inter-agent security firewall for multi-agent AI systems: runtime prompt-injection filtering, Ed25519 trust attestation, and capability enforcement for LangChain, LangGraph, CrewAI, and AutoGen. pip install inter-agent-guard

A runnable companion to the blog post *"Building E-commerce AI Agents on MongoDB."* A CrewAI shopping agent backed by MongoDB Vector Search, atomic inventory checkout with Change Streams, and (as deploy-ready artifacts) an Atlas Trigger and an Atlas Stream Processing pipeline for post-purchase intelligence.

Code samples connecting Python agentic frameworks (LangChain, LlamaIndex, CrewAI, ...) to LucidLink via the Python SDK and MCP server

AI LegalAdvisor, AI ContractsAnalyst, AI LegalStrategist, and AI Team Lead With Agno

Deterministic approval, rejection, cancellation, timeout, replay, and side-effect safety tests for AI agents—no API keys required.

EngramX — The Open-Source Agent Memory Framework. Pluggable storage, lifecycle policies, observability, and framework adapters for AI agents.

🤖 Free open-source coding agent — like Claude Code, but powered by Hugging Face's free Inference API. No paid keys, just pure AI. Built with Python, Gradio, and LLMs.

A homebrew Agent framework.

Governance harness for decision-making systems with structured review, epistemic honesty, and retrospective calibration.

Spec-first, embeddable agent harness runtime - a control plane for loop-engineered AI agents. Governed tool calls through one gateway, one normalized event stream, durable sessions.

A decoupled, zero-dependency dual-track long-term memory engine for LLM AI Agents. | 经验回想与印象整合引擎

AgentDeck SDK wraps OpenAI Agents SDK and LangGraph with sessions, streaming, events, human approval, and run control.

Multi-agent HR copilot for employee self-service, policy retrieval, compliance checks, and response synthesis.

🗣️ AgentRound — 多模型 AI 圆桌讨论框架。让多个 AI 模型围绕同一主题展开多轮辩论，支持并发调用、人类实时介入、智能 Token 裁剪、Rich 终端渲染和 Markdown 导出。兼容任意 OpenAI 兼容 API。

Generate a custom newspaper with an AI agent based on your favorite YouTube channels.

Governed, auditable AI review platform: the review and approval work a team does by hand, as a webhook-triggered LangGraph agent with human-in-the-loop approval, Langfuse tracing, an eval-gated CI pipeline, and a tamper-evident hash-chained audit trail. Provider-agnostic (mock/Anthropic/OpenAI), FastAPI, Bicep IaC for Azure Container Apps.

RSI-EAF: XRPL-grounded autonomous factory. LIVE x402 merchant — pay testnet Tag 1/2. https://published-zeta.vercel.app/pay.html

Islamic-knowledge AI service for DeenBridge — FastAPI + Gemini assistant with verified Qur'an/Hadith citations, scholar-review fallback, and guidance for Stellar/USDC payments.

MEMO, a visual memory assistant that prioritizes useful uncertainty over confident guesses.

Mystery AI Coach is a Discord AI assistant I built to help improve social confidence and communication skills. It learns from past interactions, analyzes conversations, tracks progress, awards XP for real-world actions, and provides personalized coaching and feedback. 🚀

个人财务操作系统：账户/流水/负债/应收款管理，Financial Engine 实时计算资产负债，LLM 配置中心 + DeepSeek AI 只读财务分析助手。Vue3 + FastAPI + PostgreSQL + Docker

A Claude skill that rewrites every AI response in grimdark Warhammer 40K Adeptus Mechanicus voice — 5 languages, 300+ vocabulary mappings

safetensors qwen3_5_text tool-calling function-calling agent edge-ai

gguf text-generation base_model:Kwaipilot/KAT-Coder-V2.5-Dev base_model:quantized:Kwaipilot/KAT-Coder-V2.5-Dev endpoints_compatible region:us

gguf quantized GGUF imatrix quantization imat

从 miniVERL 到 verl 0.8.0 的中文交互式实战课程：GRPO/PPO、HybridFlow、ToolAgentLoop 与 VeriCalc

Deterministic two-layer validator for LLM tool calls: stdlib structural checker in 25,000-case differential agreement with pinned jsonschema, plus the semantic contract layer where real failures live - qwen2.5-7B: 0 schema violations, 7 semantic, replayed in CI from committed data.

Open measurement infrastructure for agent-facing software. Separate operations from attempts, evidence from inference, and usage from value. Reach → Choice → Use → Utility → Value.

Open reproduction of NVIDIA's AVO paper (arXiv:2603.24517): evolutionary search where an autonomous coding agent IS the variation operator — Vary(P)=Agent(P,K,f). Runs on the Claude Code or Codex session you already have.

Ask your data. Trust the answer. — A text-to-SQL product for non-technical users.

Self-improving control kernel for AI coding agents (DAGx AGI Kernel): hard approval stops for irreversible actions, evidence-gated completion, self-learning playbook with drift governance. Behaviorally evaluated, harness-portable Agent Skill.

WordPress performance audit and optimization skills for AI coding agents — Core Web Vitals, caching layers, page builders, and managed-host constraints.

中文 AI Agent 实战手册 | Chinese AI Agent Handbook - 从零基础到全栈开发，最通俗易懂的 Agent 教程（双语 Bilingual）

AI-powered adaptive math tutor for kids (Grade 1-6) — CrewAI multi-agent pipeline with Gemini + Ollama

A personal AI crew that automates daily research tasks — monitors markets, watches YouTube channels, summarises with attribution, and delivers a morning digest email. Built on CrewAI + Claude, runs free on GitHub Actions.

AI Employee (Digital FTE) - 4-tier system replacing 10+ hrs/week of admin | Email, WhatsApp, LinkedIn, Twitter, FB, Odoo, Slack | 10 MCP Servers | Claude AI

Anthropomorphic AI Character Engine: a standardized agent template based on the SPL audit engine for fast, consistent AI character generation. Using causal anchoring and full-logic auditing, it erases OOC, emotional drift and memory loss, offering a stable foundation for AI companions, digital humans, interactive stories, smart customer service and

Guardrails for AI coding agents — scoped phases, reproducible proofs (exit 0 on the committed tree), and a bundled Multidim analysis engine. Pure Python, zero deps, local-first, Apache-2.0.

AI-powered research assistant & personal project hub

中文 Agent Skill：陪普通用户用 Hermes Agent + Weixin iLink 五步搭建自己的微信 AI 助手，可选知识库、受控 Codex 编程与日常自动化。支持 Codex / Kimi Code / Claude Code / Hermes 安装，默认最小权限。

A personal assistant with chat, long-term memory, daily digests, blog/news monitoring, and paper trading. Built for multi-channel delivery, currently running on Telegram.

static region:us

Agent skill + Python helper for the iKuai router private API — 14 domains, 105 verified func_names

Valid tool calls from any local model — a drop-in OpenAI-compatible proxy for Ollama that guarantees well-formed tool calls and restores tool_choice.

Yandex Search API plug-in for Hermes Agent

The company brain: it remembers, thinks, and does the work. Meetings become a typed graph with sources — it answers with citations, spots divergence on its own, and returns reports, automations and specs for AI executors.

Open map data for your AI — free, no account, no API key. MCP server for Overture Maps: places, geocoding, routing.

Self-hosted, MCP-native web-search backend for AI agents: SearXNG meta-search + RAG with citations + GitHub project selection + Tavily-compatible API. Free & local.

Mail for AI agents ( and humans). MCP server giving Claude (or any agent) safe, permission-controlled access to email (Microsoft Graph + IMAP), calendar and your knowledge files. Two-phase confirmation for sends, audit log, local index. No built-in LLM: your agent brings the intelligence.

LittleOrange Video MCP 是基于 Model Context Protocol (MCP) 的视频生成服务端，为 AI 应用提供统一、标准化的视频生成 API 调用能力。服务基于官方视频生成 API 文档封装，支持 Sora2、Veo3.1、Vidu Q3、Dreamina Seedance 2.0 等多个主流视频生成模型，内置智能异步轮询、三层工具架构（底层API映射、自动轮询工具、高层Agent工具）、完整素材管理等功能，完美适配各类 AI 客户端与 Agent 场景。

EcoHub AI is a full-stack outdoor social platform that helps users discover scenic spots worldwide, organize meetups, join outdoor communities, chat in real-time, and generate personalized travel plans — all powered by AI.

docsforge — Universal software documentation → Markdown for LLMs

AI-powered IDA Pro REST API with 90+ endpoints. Autonomous RE agent, swarm worker, full SDK coverage.

Database-to-AI automation: reverse-engineer, package as REST microservice, connect to AI agents via MCP.

Answers "where should I submit this paper?" from cited soft metadata — reviewer culture, framing requirements, real word limits — instead of model recall. 399 curated journal entries, a 166,821-journal ISSN spine, and blanks where nobody actually knows.

gguf text-generation base_model:01-ai/Yi-1.5-34B-Chat base_model:quantized:01-ai/Yi-1.5-34B-Chat license:other region:us

transformers safetensors qwen2 text-generation code codeqwen

transformers gguf code codeqwen chat qwen

Local A-share research, LLM decisioning, deterministic risk control, and daily simulation

Open-source financial research agent whose false-positive rate is measured, not assumed. Permutation tests, multiple-testing correction, and a published calibration study that found (and fixed) a 45% false-positive rate.

终端里的编码 agent：接任意 OpenAI 兼容模型，任务计划/子代理/长期记忆/断网沙盒

An open-source AI agronomy agent that runs locally on your own machine or server, reaches you on Telegram and WhatsApp, and computes aquaponics system designs from a deterministic, source-cited engineering core instead of guessing them.

A headless jadx engine for LLM agents. Answers structured questions about an APK, DEX or JAR over a plain CLI. No GUI, no daemon.

Principal-level Hermes Agent profile and skill pack for production RAG, AI agents, evaluation, observability, and implementation-ready architecture docs.

mlx safetensors qwen3_5_moe apple-silicon agentworld 5bit

gradio region:us

streamlit region:us

streamlit region:us

gradio region:us

streamlit region:us

streamlit region:us

streamlit region:us

streamlit region:us

streamlit region:us

streamlit region:us

streamlit region:us

streamlit region:us