Skip to content

LangChain/LangGraph vs Hermes

LangChain, LangGraph và Hermes có liên quan, nhưng không cùng vai trò.

ToolTầngOutput chính
LangChainAgent app frameworkAI application hoặc agent logic
LangGraphStateful agent orchestration framework/runtimeLong-running graph-based agent service
HermesAgent harness/runtime CLIRunning/customizable agent với tools, memory, skills, subagents

Phân biệt đơn giản

mermaid
flowchart TB
    A[Build an AI app or agent backend] --> B[LangChain / LangGraph]
    B --> C[Agent service]

    D[Run a coding/research agent harness] --> E[Hermes]
    E --> F[Repo changes / tool execution]

    G[Need disciplined delivery workflow] --> H[OpenSpec / Spec Kit / AI-DLC / Superpowers]

LangGraph vs Hermes

Câu hỏiLangGraphHermes
Là gì?Framework/runtime để build stateful agent appsOpen-source agent CLI/runtime
Ai dùng?Developers build agent backendsDevelopers/platform teams chạy/customize agents
Bạn viết gì?Graph state, nodes, edges, toolsRuntime config, tools, skills, workflow instructions
OutputAgent service/applicationRunning agent harness
Phù hợp nhấtLong-running stateful app workflowsHackable coding/research agent execution

Có kết hợp được không?

Có. Các combo thường gặp:

ComboÝ nghĩa
LangGraph + OpenSpecOpenSpec govern changes cho LangGraph app
LangGraph + AI-DLCAI-DLC govern high-risk agent app delivery
Hermes + LangGraphHermes là coding/runtime harness; LangGraph là app framework đang được build
LangChain + SuperpowersImplement LangChain app với TDD/review discipline

Đừng nhầm điều này

Đừng dùng LangGraph để thay delivery governance.

text
LangGraph orchestrates runtime behavior.
AI-DLC governs delivery decisions.
OpenSpec/Spec Kit manage specs.
Superpowers enforces coding discipline.
Hermes runs/customizes agent execution.

Stack ví dụ

mermaid
flowchart TB
    A[OpenSpec change proposal] --> B[Developer uses Hermes or Codex CLI]
    B --> C[Implement LangGraph agent]
    C --> D[LangGraph service]
    D --> E[Tests and evals]
    E --> F[Sync/archive OpenSpec change]

Built as a static bilingual AI engineering stack guide.