mirror of
https://github.com/lobehub/lobehub.git
synced 2026-09-20 04:56:13 +08:00
YuTengjing
191f3fe776
* 📝 docs: fix outdated content across the docs site Audited docs/ against the current codebase and fixed stale content that drifted after major refactors (Clerk/NextAuth → Better Auth, src/server → apps/server, src/{config,envs,locales} → packages/*, model-bank updates). - self-hosting/auth: mark removed NextAuth/Clerk pages with deprecation warnings; fix microsoft/zitadel alias claims and the Logto callback path - environment-variables & advanced: correct wrong defaults (S3_SET_ACL, AI_IMAGE_DEFAULT_IMAGE_NUM, index/proxy URLs), drop non-existent env vars and fictional feature flags, fix OIDC config guidance - development: update refactored paths, Better Auth, .mjs configs, remove the MSW section (msw is not a dependency) - providers & model examples: replace retired model ids with in-bank ones; fix comfyui frontmatter corruption - usage: rename Moderator→Orchestrator, GTD→Task, correct shortcut keys, channel capability tables, Feishu/Lark WebSocket mode - platform/migration: fix dead links, docker service name, npx/npm → bun EN and zh-CN pairs updated together. * 📝 docs: fix Logto callback path and AUTH_SECRET deprecation guidance - Revert Logto/Sealos callback URLs to /api/auth/callback/logto to match the runtime SSO redirectURI override - Narrow legacy auth Callout: AUTH_SECRET remains required for Better Auth; mark NEXT_AUTH_SECRET as deprecated in zh-CN * 📝 docs: refresh agent skills for locales, env, auth, and agentRun paths Align OSS agent guidance with post-refactor layout: - packages/locales/src/default + bun run i18n - packages/env/src for env schemas (@/envs alias) - spa-auth shell + src/routes/auth segments - chat agentRun/transports/hetero instead of aiChat * 📝 docs: document APP_URL as required for Better Auth base URL Self-hosted auth uses appEnv.APP_URL as baseURL, not request-header auto-detection. NEXT_PUBLIC_AUTH_URL remains deprecated. * 📝 docs: note self-host search provider setup for web search Built-in lobe-web-browsing needs no client plugin, but self-hosted servers still require SEARCH_PROVIDERS / SEARXNG_URL (default SearXNG). * 📝 docs: point feature guide welcome UI at AgentHome Opening message/questions on the agent chat page live in src/features/AgentHome, not Onboarding WelcomeMessage.
142 lines
7.0 KiB
Plaintext
142 lines
7.0 KiB
Plaintext
---
|
||
title: Agent Groups
|
||
description: >-
|
||
Bring multiple specialized Agents together to collaborate — sequential,
|
||
parallel, iterative, or debate mode.
|
||
tags:
|
||
- LobeHub
|
||
- Agent Group
|
||
- Multi-Agent Collaboration
|
||
- Group Chat
|
||
- Orchestrator
|
||
- Sequential Mode
|
||
- Parallel Mode
|
||
---
|
||
|
||
# Agent Groups
|
||
|
||
One Agent's perspective is rarely enough. Complex problems need multiple angles; creative projects thrive on diverse expertise; decisions benefit from structured debate. Agent Group Chat assembles a team of specialized Agents that work together like a real group — a research Agent, an analysis Agent, and a writing Agent each contributing their strengths to your task. You're not just getting an answer; you're running a coordinated process.
|
||
|
||
Pose a question or task, and each Agent responds from its area of expertise. They can build on each other's outputs, challenge assumptions, or work in parallel — depending on the mode you choose.
|
||
|
||
## Why Not Just One Agent?
|
||
|
||
- A single Agent can only analyze from one perspective
|
||
- Expertise is bounded by a single predefined role
|
||
- There's no back-and-forth to surface blind spots or trade-offs
|
||
|
||
Agent Groups fix all three.
|
||
|
||
## Collaboration Modes
|
||
|
||
Agent Groups adapt to the task:
|
||
|
||
**Sequential** — Agents work one after another in a defined order. Example: Research Agent gathers information → Analysis Agent processes data → Writing Agent creates the final document. Best for linear workflows with clear handoff points.
|
||
|
||
**Parallel** — Multiple Agents tackle different aspects simultaneously. Example: Marketing Agent writes copy while Data Agent compiles metrics. Best for independent subtasks.
|
||
|
||
**Iterative** — Agents review and refine work through multiple rounds. Example: Writer creates draft → Editor provides feedback → Writer revises → Editor does final review. Best for high-quality output that requires polish.
|
||
|
||
**Debate** — Agents argue different positions. An advocate makes the case, a critic challenges assumptions, an analyst weighs evidence, and a mediator synthesizes conclusions. Best for complex decisions that benefit from structured disagreement.
|
||
|
||
## About the Orchestrator
|
||
|
||
Every Agent Group includes a built-in Orchestrator responsible for:
|
||
|
||
- Understanding your goal and assigning tasks to the right Agents
|
||
- Coordinating the order of contributions
|
||
- Summarizing the discussion and extracting key conclusions
|
||
- Keeping the conversation organized and on-topic
|
||
|
||
## Create an Agent Group
|
||
|
||
Click **Create Group** in the left sidebar to get started.
|
||
|
||
|
||
|
||
When creating a group, you can start from an existing template or assemble your own team. Choose whether to include an Orchestrator and select the model for it.
|
||
|
||
|
||
|
||
## Configure an Agent Group
|
||
|
||
In the group chat session, select an Agent in the left sidebar to swap its model or remove it from the group.
|
||
|
||
|
||
|
||
Click **Add Member** in the left sidebar to bring additional Agents into the group.
|
||
|
||
|
||
|
||
Go to **Group Profile** in the left sidebar to edit the group prompt, add Skills, or change the Orchestrator model. You can also use Agent Builder on the right panel — describe your goal and it will generate the complete group configuration, including settings, system prompts, and Skill setup.
|
||
|
||
|
||
|
||
## Example Agent Groups
|
||
|
||
### Content Creation Team
|
||
|
||
A sequential workflow for producing blog posts or articles:
|
||
|
||
1. **Researcher** — Gathers information, credible sources, statistics, and expert quotes
|
||
2. **Writer** — Writes an engaging draft using the research material
|
||
3. **Editor** — Reviews and refines for clarity, flow, and consistency
|
||
4. **SEO Specialist** — Optimizes keywords, headings, and meta description
|
||
|
||
### Research Analysis Team
|
||
|
||
1. **Data Collector** — Gathers data from multiple sources and documents
|
||
2. **Statistician** — Analyzes data, calculates metrics, identifies patterns
|
||
3. **Domain Expert** — Interprets findings in the context of industry knowledge
|
||
4. **Report Writer** — Synthesizes insights into an executive summary
|
||
|
||
### Software Development Team
|
||
|
||
- **Requirements Agent** — Gathers needs and writes product requirements documents
|
||
- **Technical Architect** — Designs system architecture and technical specifications
|
||
- **Developer** — Writes implementation code and technical documentation
|
||
- **QA Tester** — Reviews code, suggests test cases, identifies edge cases
|
||
|
||
## Best Practices
|
||
|
||
**Define clear roles** — Each Agent should have a specific, well-defined responsibility:
|
||
|
||
- ❌ Vague: "Agent 1: Help with content"
|
||
- ✅ Clear: "Research Agent: Find 3–5 credible sources on the topic and summarize key points. Include statistics and expert quotes."
|
||
|
||
**Avoid redundancy** — Don't create multiple Agents with overlapping roles. Each Agent should bring unique value.
|
||
|
||
**Set clear handoff points** — In sequential workflows, define exactly what each Agent passes to the next. Example: "Research Agent outputs a Markdown document with sources, key points, and data. Writer Agent uses this as input to draft."
|
||
|
||
**Right-size the team** — More Agents does not mean better results. Start with 2–3 Agents and add more only as needed. Too many Agents slow the workflow without adding value.
|
||
|
||
**Match models to tasks** — Use capable models where reasoning matters; use faster, lighter models for straightforward steps.
|
||
|
||
## Common Use Cases
|
||
|
||
**Content production** — Blog posts (research → write → edit → SEO), social media campaigns, documentation, video scripts.
|
||
|
||
**Business analysis** — Market research, competitive analysis, financial modeling, risk assessment.
|
||
|
||
**Software development** — Feature development (requirements → design → code → testing), code review, bug fixing, API design.
|
||
|
||
**Creative projects** — Story development, marketing campaigns, product naming, design critique.
|
||
|
||
## Troubleshooting
|
||
|
||
**Group responds too slowly** — Reduce the number of Agents, use lighter models for simple steps, switch to parallel mode when possible, or simplify Agent system roles.
|
||
|
||
**Inconsistent outputs** — Add clearer role definitions, include shared guidelines in the group context, add a reviewer Agent, or lower the temperature setting.
|
||
|
||
**Agents not collaborating well** — Explicitly define what each Agent receives and produces, add transition instructions between Agents, or switch to sequential mode.
|
||
|
||
**Poor quality results** — Upgrade to a better model for critical Agents, add specialist Agents for key areas, include a quality-control Agent, or write more detailed system roles.
|
||
|
||
<Cards>
|
||
<Card href={'/docs/usage/getting-started/agent'} title={'Agent'} />
|
||
|
||
<Card href={'/docs/usage/agent/scheduled-task'} title={'Scheduled Tasks'} />
|
||
|
||
<Card href={'/docs/usage/getting-started/page'} title={'Pages'} />
|
||
</Cards>
|