AIエージェント
agent-rules-books
Battle-tested coding rules for AI agents, distilled from classic programming books.
★ 2.9k スターMIT
ガイドを見る →parlantは、emcie-coが公開しているオープンソースのAIエージェントプロジェクトです（GitHub: emcie-co/parlant）。GitHubスターは18,301、フォークは1,554で、Apache-2.0ライセンスの下で公開されています。GitHubでの最終更新は2026年7月12日です。
Build customer-facing AI agents that actually follow your rules
List the rules, journeys and approved responses your customer agent must follow.
Install the parlant package from PyPI.
Express behavior as guidelines, relationships and journeys instead of prompt text.
Start the agent and test a customer scenario, including an edge case it must handle carefully.
Behavioral rules are defined in code and matched to each turn, so adding rules makes the agent smarter, not more confused.
Model multi-step conversational flows with branches that the engine keeps the agent on.
At critical moments, restrict the agent to pre-approved response templates so wording stays exact.
Define domain terms and rule priorities so conflicts resolve the way you intend.
Attach tools the agent can call while the harness keeps interactions controlled and predictable.
Inspect why the agent acted as it did, and run it with the major LLM providers.
Get the source code from GitHub and enter the project directory.
git clone https://github.com/emcie-co/parlant.git
cd parlantAdd the package to your Python project.
pip install parlantThe README points to a 5-minute quickstart that builds a first controlled agent.
# Follow the 5-minute quickstart linked from the READMEIt is maintained by emcie-co and contributors on GitHub.
Yes - Parlant is Apache-2.0 licensed and free to use.
Parlant is a harness, not a model: you connect your own LLM (OpenAI, Gemini, open models and others per the docs), so you need that provider's key or a local model.
Compared with LangGraph or DSPy, Parlant is specialized for conversational control - guidelines, journeys and canned responses for customer-facing reliability - rather than general agent orchestration or prompt optimization.
Run pip install parlant and complete the 5-minute quickstart, then add one real guideline from your support playbook.
Battle-tested coding rules for AI agents, distilled from classic programming books.
Build, deploy, and monetize autonomous agents across Web3.
Upsonic — build autonomous AI agents in Python, with prebuilt agents, custom tools, MCP and OCR built in.
Build production-ready agentic workflows from natural language — agents that operate browsers and computers.
Sync your AI coding setup: skills, agents, rules, MCP, hooks, everywhere.
Build, run, and scale AI agents like APIs and microservices.
LibreAIカタログに掲載されたAIエージェントプロジェクトの中で最もスターの多いもの — parlantが合わなかった場合に検討したい、現実的な候補リストです。
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
カタログの中でparlantとスター数が近いAIエージェントプロジェクト — 前後の棚もそのまま見て回れます。
A self-healing harness that connects an LLM directly to your real browser
The JavaScript/TypeScript version of LangChain - the agent engineering platform
Microsoft's trainer for AI agents - reinforcement learning for almost any agent framework
An open-source agent operating system in Rust - autonomous 'Hands' that do real work
Unit tests for LLMs — evaluate before you ship.
A production-grade platform for building agentic chat bots on every IM platform