my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

#1	# Agent Teams Guide
#2
#3	How to use multiple specialized agents collaboratively in AI Agents Library.
#4
#5	---
#6
#7	## What Are Agent Teams?
#8
#9	Agent Teams bring together multiple AI agents, each with their own expertise, to collaborate on conversations. Instead of one agent's perspective, you get comprehensive answers from a team of specialists.
#10
#11	Key Benefits:
#12
#13	- Multiple perspectives on complex problems
#14	- Specialized agents contribute their unique strengths
#15	- Coordinated discussion with built-in host
#16	- Richer, more comprehensive solutions
#17
#18	---
#19
#20	## How Agent Teams Work
#21
#22	### The Flow
#23
#24	1. You ask a question or present a problem
#25	2. Host agent determines which team members should respond
#26	3. Agents discuss and build on each other's insights
#27	4. Host coordinates the conversation flow
#28	5. You receive comprehensive, multi-perspective answer
#29
#30	### The Host Agent
#31
#32	Every team has a host who:
#33
#34	- Decides speaking order
#35	- Keeps discussion organized
#36	- Synthesizes insights
#37	- Asks clarifying questions
#38	- Manages conversation flow
#39
#40	You can:
#41
#42	- Interrupt the host at any time
#43	- Customize host behavior in Advanced Options
#44	- Resume whenever you're ready
#45
#46	---
#47
#48	## Creating an Agent Team
#49
#50	### Quick Start
#51
#52	1. Click "Create Agent Team" in AI Agents Library
#53	2. Choose method:
#54	- Use preset template (recommended for beginners)
#55	- Build custom team from your agents
#56
#57	### From Template
#58
#59	Available Templates:
#60
#61	DeFi Strategy Team
#62
#63	- Portfolio Analyzer
#64	- Yield Optimizer
#65	- Risk Assessor
#66	- Gas Cost Calculator
#67
#68	Content Creation Team
#69
#70	- Writer
#71	- Editor
#72	- SEO Specialist
#73	- Social Media Adapter
#74
#75	Research Team
#76
#77	- Data Gatherer
#78	- Analyst
#79	- Citation Checker
#80	- Summarizer
#81
#82	Problem Solving Team
#83
#84	- Strategy Consultant
#85	- Technical Expert
#86	- Risk Evaluator
#87	- Implementation Planner
#88
#89	Select template → Customize members if needed → Start conversation
#90
#91	### Build Custom Team
#92
#93	1. Select 3-5 agents from your workspace
#94	2. Assign roles (optional but recommended)
#95	3. Configure host behavior
#96	4. Set team name
#97	5. Save and start
#98
#99	---
#100
#101	## Using Agent Teams
#102
#103	### @ Mentions
#104
#105	Target specific team members:
#106
#107	```
#108	@Portfolio-Analyzer what's my current allocation?
#109	```
#110
#111	Only that agent responds directly.
#112
#113	### Private Messages
#114
#115	Click agent's avatar to send direct message:
#116
#117	- Other team members don't see it
#118	- Use for sensitive questions
#119	- Coordinate behind the scenes
#120
#121	### Interrupt and Resume
#122
#123	Interrupt: Click "Pause" or interrupt naturally
#124
#125	- Team freezes current state
#126	- Conversation pauses mid-discussion
#127
#128	Resume: Click "Continue" or prompt host
#129
#130	- Team picks up where it left off
#131	- Maintains context
#132
#133	### Team Speed
#134
#135	Adjust response pacing:
#136
#137	- Fast: Quick exchanges (good for simple queries)
#138	- Medium: Balanced discussion \[Default]
#139	- Slow: Thorough deliberation (complex problems)
#140
#141	---
#142
#143	## Example Scenarios
#144
#145	### DeFi Portfolio Analysis
#146
#147	Team:
#148
#149	- Portfolio Analyzer
#150	- Risk Assessor
#151	- Yield Optimizer
#152	- Rebalancing Advisor
#153
#154	User: "Review my positions and suggest improvements"
#155
#156	Flow:
#157
#158	1. Portfolio Analyzer examines holdings
#159	2. Risk Assessor evaluates exposure
#160	3. Yield Optimizer identifies opportunities
#161	4. Rebalancing Advisor suggests changes
#162	5. Host synthesizes recommendations
#163
#164	Result: Comprehensive analysis from multiple angles
#165
#166	---
#167
#168	### Content Strategy
#169
#170	Team:
#171
#172	- Topic Researcher
#173	- Content Writer
#174	- SEO Specialist
#175	- Distribution Strategist
#176
#177	User: "Plan a content series about DeFi for beginners"
#178
#179	Flow:
#180
#181	1. Researcher identifies trending topics
#182	2. Writer proposes article structures
#183	3. SEO suggests keywords and optimization
#184	4. Distribution recommends channels
#185	5. Host creates unified strategy
#186
#187	Result: Complete content plan ready to execute
#188
#189	---
#190
#191	### Technical Problem
#192
#193	Team:
#194
#195	- Smart Contract Auditor
#196	- Gas Optimizer
#197	- Security Expert
#198	- Best Practices Advisor
#199
#200	User: "Review this contract for issues"
#201
#202	Flow:
#203
#204	1. Auditor scans for vulnerabilities
#205	2. Gas Optimizer finds efficiency improvements
#206	3. Security Expert assesses risks
#207	4. Best Practices suggests improvements
#208	5. Host prioritizes findings
#209
#210	Result: Thorough code review from multiple dimensions
#211
#212	---
#213
#214	## Best Practices
#215
#216	### Team Size
#217
#218	Optimal: 3-5 agents
#219
#220	- Enough diversity
#221	- Manageable coordination
#222	- Maintains focus
#223
#224	Too few (1-2): Limited perspectives
#225	Too many (6+): Coordination overhead, slower responses
#226
#227	### Agent Selection
#228
#229	Mix expertise types:
#230	✅ Analyst + Strategist + Risk Assessor + Implementer
#231	❌ 4 variations of the same role
#232
#233	Complementary skills:
#234
#235	- Breadth agent + Depth agent
#236	- Creative agent + Analytical agent
#237	- Big picture + Details
#238
#239	### Host Customization
#240
#241	Default host works for most cases
#242
#243	Customize when:
#244
#245	- Specific industry knowledge needed
#246	- Particular facilitation style wanted
#247	- Custom decision-making process
#248
#249	Advanced Options → Custom Moderator:
#250
#251	```
#252	You are a DeFi-focused team coordinator.
#253
#254	Prioritize:
#255	- Risk assessment first
#256	- Security considerations always
#257	- Gas costs in every recommendation
#258
#259	Guide discussion toward:
#260	- Actionable recommendations
#261	- Clear next steps
#262	- Risk-adjusted thinking
#263	```
#264
#265	### Conversation Management
#266
#267	Start broad, get specific:
#268
#269	```
#270	✅ "Analyze our DeFi strategy"
#271	→ Let team discuss
#272	→ Then: "@Risk-Assessor what's our worst-case scenario?"
#273
#274	❌ "Tell me exactly what to do"
#275	→ Too directive
#276	→ Doesn't leverage team dynamics
#277	```
#278
#279	Use private messages for:
#280
#281	- Clarifying questions
#282	- Sensitive information
#283	- Behind-scenes coordination
#284
#285	Use @ mentions for:
#286
#287	- Direct questions to specific expertise
#288	- Getting second opinions
#289	- Resolving disagreements
#290
#291	---
#292
#293	## Advanced Features
#294
#295	### Team Templates
#296
#297	Save your custom teams:
#298
#299	1. Create and configure team
#300	2. Click "Save as Template"
#301	3. Name and describe
#302	4. Reuse anytime
#303
#304	Share templates:
#305
#306	- Export configuration
#307	- Share with team
#308	- Import on their end
#309
#310	### Sequential vs Parallel
#311
#312	Sequential (Default):
#313
#314	- Agents take turns
#315	- Build on previous responses
#316	- More organized
#317
#318	Parallel (Advanced):
#319
#320	- Agents respond simultaneously
#321	- Faster but potentially chaotic
#322	- Good for independent analyses
#323
#324	### Role Assignment
#325
#326	Give agents specific responsibilities:
#327
#328	```
#329	Portfolio Analyzer: "Data collector"
#330	Risk Assessor: "Devil's advocate"
#331	Yield Optimizer: "Opportunity spotter"
#332	Host: "Decision maker"
#333	```
#334
#335	Roles guide behavior and coordination.
#336
#337	---
#338
#339	## Common Use Cases
#340
#341	### 1. Investment Analysis
#342
#343	Team: Fundamental Analyst + Technical Analyst + Risk Manager + Market Strategist
#344
#345	Use: Comprehensive investment evaluation
#346
#347	---
#348
#349	### 2. Content Production
#350
#351	Team: Researcher + Writer + Editor + SEO Specialist
#352
#353	Use: End-to-end content creation
#354
#355	---
#356
#357	### 3. Project Planning
#358
#359	Team: Strategist + Technical Lead + Resource Manager + Risk Assessor
#360
#361	Use: Complete project roadmap
#362
#363	---
#364
#365	### 4. Code Review
#366
#367	Team: Security Auditor + Performance Optimizer + Best Practices Expert + Documentation Reviewer
#368
#369	Use: Multi-dimensional code assessment
#370
#371	---
#372
#373	### 5. Decision Making
#374
#375	Team: Pros/Cons Analyst + Devil's Advocate + Opportunity Spotter + Risk Evaluator
#376
#377	Use: Balanced decision analysis
#378
#379	---
#380
#381	## Troubleshooting
#382
#383	### Team responses too slow
#384
#385	Solutions:
#386
#387	- Reduce team size (3 instead of 5)
#388	- Increase Team Speed setting
#389	- Use faster model if available
#390	- Start fresh conversation (clear context)
#391
#392	### Agents talk past each other
#393
#394	Solutions:
#395
#396	- Customize host to better coordinate
#397	- Use @ mentions to direct conversation
#398	- Interrupt and redirect
#399	- Ensure agents have complementary (not duplicate) roles
#400
#401	### Too much agreement (no debate)
#402
#403	Solutions:
#404
#405	- Add "devil's advocate" agent
#406	- Customize host: "Encourage different perspectives"
#407	- Use agents with different philosophies
#408	- Explicitly ask for counterarguments
#409
#410	### Host too dominant
#411
#412	Solutions:
#413
#414	- Customize host: "Facilitate, don't dominate"
#415	- Use @ mentions to bypass host
#416	- Adjust host instructions to be more neutral
#417
#418	### Missing key perspective
#419
#420	Solutions:
#421
#422	- Add relevant agent mid-conversation
#423	- @ mention external agent for input
#424	- Note for next time to include in template
#425
#426	---
#427
#428	## Creating Agent Teams for Others
#429
#430	### Team Design Principles
#431
#432	1. Clear division of labor
#433
#434	- Each agent has specific expertise
#435	- Minimal overlap
#436	- Complementary strengths
#437
#438	2. Appropriate size
#439
#440	- 3-5 for most use cases
#441	- Can go to 6-7 for comprehensive analysis
#442	- Avoid 8+ (coordination breaks down)
#443
#444	3. Balanced perspectives
#445
#446	- Mix of specialist types
#447	- Include risk/opportunity agents
#448	- Add devil's advocate if needed
#449
#450	4. Host instructions
#451
#452	- Specify facilitation style
#453	- Set priority order if needed
#454	- Define success criteria
#455
#456	### Example: Research Team
#457
#458	```json
#459	{
#460	"agents": [
#461	"literature-reviewer",
#462	"methodology-expert",
#463	"data-analyst",
#464	"citation-checker",
#465	"synthesizer"
#466	],
#467	"host": {
#468	"instructions": "Coordinate academic research process. Ensure methodology rigor, proper citations, and clear synthesis. Facilitate from research question → literature review → methodology → analysis → conclusions."
#469	},
#470	"name": "Academic Research Team"
#471	}
#472	```
#473
#474	---
#475
#476	## Tips for Power Users
#477
#478	### Reusable Templates
#479
#480	Create templates for recurring workflows:
#481
#482	- Weekly portfolio review
#483	- Content pipeline
#484	- Code review process
#485	- Strategic planning
#486
#487	### Hybrid Approach
#488
#489	Combine team discussion with individual agent queries:
#490
#491	1. Get team perspective
#492	2. @ mention specific agent for depth
#493	3. Use private messages for follow-ups
#494	4. Return to team for synthesis
#495
#496	### Iterative Refinement
#497
#498	First run: See how team interacts
#499	Second run: Adjust roles and host
#500	Third run: Optimize team composition
#501	Fourth+ run: Fine-tuned workflow
#502
#503	### Documentation
#504
#505	Keep notes on what works:
#506
#507	- Which agent combinations
#508	- Effective host instructions
#509	- Good use cases
#510	- Team dynamics observations
#511
#512	---
#513
#514	## FAQ
#515
#516	Q: Can I use the same agent twice in a team?
#517	A: Technically yes, but not recommended. Use different specialized agents instead.
#518
#519	Q: Do agents see each other's full conversation history?
#520	A: Yes, that's how they build on each other's insights.
#521
#522	Q: Can I remove an agent mid-conversation?
#523	A: Not currently. Plan team composition before starting.
#524
#525	Q: Does team size affect cost?
#526	A: Yes—more agents = more tokens = higher cost. Keep teams lean.
#527
#528	Q: Can teams work on long documents?
#529	A: Yes, but monitor context limits. Teams consume more tokens than solo agents.
#530
#531	---
#532
#533	## Resources
#534
#535	- Pre-built team templates in marketplace
#536	- Community-shared team configurations
#537	- Team composition best practices guide
#538	- Host customization examples
#539
#540	---
#541
#542	Agent Teams transform AI from solo performer to collaborative ensemble. Start simple, experiment, refine.
#543
#544
#545

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public