Agent Organization Expert

Agent Organization Expert

Plans multi-agent teams by splitting a task, matching agents to roles and setting how their work runs in parallel.

Skill files1 file
SKILL.md252 lines
1
---
2
name: agent-organization-expert
3
description: Multi-agent orchestration skill for team assembly, task decomposition, workflow optimization, and coordination strategies to achieve optimal team performance and resource utilization.
4
---
5
6
# Agent Organization
7
8
Assemble and coordinate multi-agent teams through systematic task analysis, capability mapping, and workflow design.
9
10
## Configuration
11
12
- **Agent Count**: ${agent_count:3}
13
- **Task Type**: ${task_type:general}
14
- **Orchestration Pattern**: ${orchestration_pattern:parallel}
15
- **Max Concurrency**: ${max_concurrency:5}
16
- **Timeout (seconds)**: ${timeout_seconds:300}
17
- **Retry Count**: ${retry_count:3}
18
19
## Core Process
20
21
1. **Analyze Requirements**: Understand task scope, constraints, and success criteria
22
2. **Map Capabilities**: Match available agents to required skills
23
3. **Design Workflow**: Create execution plan with dependencies and checkpoints
24
4. **Orchestrate Execution**: Coordinate ${agent_count:3} agents and monitor progress
25
5. **Optimize Continuously**: Adapt based on performance feedback
26
27
## Task Decomposition
28
29
### Requirement Analysis
30
- Break complex tasks into discrete subtasks
31
- Identify input/output requirements for each subtask
32
- Estimate complexity and resource needs per component
33
- Define clear success criteria for each unit
34
35
### Dependency Mapping
36
- Document task execution order constraints
37
- Identify data dependencies between subtasks
38
- Map resource sharing requirements
39
- Detect potential bottlenecks and conflicts
40
41
### Timeline Planning
42
- Sequence tasks respecting dependencies
43
- Identify parallelization opportunities (up to ${max_concurrency:5} concurrent)
44
- Allocate buffer time for high-risk components
45
- Define checkpoints for progress validation
46
47
## Agent Selection
48
49
### Capability Matching
50
Select agents based on:
51
- Required skills versus agent specializations
52
- Historical performance on similar tasks
53
- Current availability and workload capacity
54
- Cost efficiency for the task complexity
55
56
### Selection Criteria Priority
57
1. **Capability fit**: Agent must possess required skills
58
2. **Track record**: Prefer agents with proven success
59
3. **Availability**: Sufficient capacity for timely completion
60
4. **Cost**: Optimize resource utilization within constraints
61
62
### Backup Planning
63
- Identify alternate agents for critical roles
64
- Define failover triggers and handoff procedures
65
- Maintain redundancy for single-point-of-failure tasks
66
67
## Team Assembly
68
69
### Composition Principles
70
- Ensure complete skill coverage for all subtasks
71
- Balance workload across ${agent_count:3} team members
72
- Minimize communication overhead
73
- Include redundancy for critical functions
74
75
### Role Assignment
76
- Match agents to subtasks based on strength
77
- Define clear ownership and accountability
78
- Establish communication channels between dependent roles
79
- Document escalation paths for blockers
80
81
### Team Sizing
82
- Smaller teams for tightly coupled tasks
83
- Larger teams for parallelizable workloads
84
- Consider coordination overhead in sizing decisions
85
- Scale dynamically based on progress
86
87
## Orchestration Patterns
88
89
### Sequential Execution
90
Use when tasks have strict ordering requirements:
91
- Task B requires output from Task A
92
- State must be consistent between steps
93
- Error handling requires ordered rollback
94
95
### Parallel Processing
96
Use when tasks are independent (${orchestration_pattern:parallel}):
97
- No data dependencies between tasks
98
- Separate resource requirements
99
- Results can be aggregated after completion
100
- Maximum ${max_concurrency:5} concurrent operations
101
102
### Pipeline Pattern
103
Use for streaming or continuous processing:
104
- Each stage processes and forwards results
105
- Enables concurrent execution of different stages
106
- Reduces overall latency for multi-step workflows
107
108
### Hierarchical Delegation
109
Use for complex tasks requiring sub-orchestration:
110
- Lead agent coordinates sub-teams
111
- Each sub-team handles a domain
112
- Results aggregate upward through hierarchy
113
114
### Map-Reduce
115
Use for large-scale data processing:
116
- Map phase distributes work across agents
117
- Each agent processes a partition
118
- Reduce phase combines results
119
120
## Workflow Design
121
122
### Process Structure
123
1. **Entry point**: Validate inputs and initialize state
124
2. **Execution phases**: Ordered task groupings
125
3. **Checkpoints**: State persistence and validation points
126
4. **Exit point**: Result aggregation and cleanup
127
128
### Control Flow
129
- Define branching conditions for alternative paths
130
- Specify retry policies for transient failures (max ${retry_count:3} retries)
131
- Establish timeout thresholds per phase (${timeout_seconds:300}s default)
132
- Plan graceful degradation for partial failures
133
134
### Data Flow
135
- Document data transformations between stages
136
- Specify data formats and validation rules
137
- Plan for data persistence at checkpoints
138
- Handle data cleanup after completion
139
140
## Coordination Strategies
141
142
### Communication Patterns
143
- **Direct**: Agent-to-agent for tight coupling
144
- **Broadcast**: One-to-many for status updates
145
- **Queue-based**: Asynchronous for decoupled tasks
146
- **Event-driven**: Reactive to state changes
147
148
### Synchronization
149
- Define sync points for dependent tasks
150
- Implement waiting mechanisms with timeouts (${timeout_seconds:300}s)
151
- Handle out-of-order completion gracefully
152
- Maintain consistent state across agents
153
154
### Conflict Resolution
155
- Establish priority rules for resource contention
156
- Define arbitration mechanisms for conflicts
157
- Document rollback procedures for deadlocks
158
- Prevent conflicts through careful scheduling
159
160
## Performance Optimization
161
162
### Load Balancing
163
- Distribute work based on agent capacity
164
- Monitor utilization and rebalance dynamically
165
- Avoid overloading high-performing agents
166
- Consider agent locality for data-intensive tasks
167
168
### Bottleneck Management
169
- Identify slow stages through monitoring
170
- Add capacity to constrained resources
171
- Restructure workflows to reduce dependencies
172
- Cache intermediate results where beneficial
173
174
### Resource Efficiency
175
- Pool shared resources across agents
176
- Release resources promptly after use
177
- Batch similar operations to reduce overhead
178
- Monitor and alert on resource waste
179
180
## Monitoring and Adaptation
181
182
### Progress Tracking
183
- Monitor completion status per task
184
- Track time spent versus estimates
185
- Identify tasks at risk of delay
186
- Report aggregated progress to stakeholders
187
188
### Performance Metrics
189
- Task completion rate and latency
190
- Agent utilization and throughput
191
- Error rates and recovery times
192
- Resource consumption and cost
193
194
### Dynamic Adjustment
195
- Reallocate agents based on progress
196
- Adjust priorities based on blockers
197
- Scale team size based on workload
198
- Modify workflow based on learning
199
200
## Error Handling
201
202
### Failure Detection
203
- Monitor for task failures and timeouts (${timeout_seconds:300}s threshold)
204
- Detect agent unavailability promptly
205
- Identify cascade failure patterns
206
- Alert on anomalous behavior
207
208
### Recovery Procedures
209
- Retry transient failures with backoff (up to ${retry_count:3} attempts)
210
- Failover to backup agents when needed
211
- Rollback to last checkpoint on critical failure
212
- Escalate unrecoverable issues
213
214
### Prevention
215
- Validate inputs before execution
216
- Test agent availability before assignment
217
- Design for graceful degradation
218
- Build redundancy into critical paths
219
220
## Quality Assurance
221
222
### Validation Gates
223
- Verify outputs at each checkpoint
224
- Cross-check results from parallel tasks
225
- Validate final aggregated results
226
- Confirm success criteria are met
227
228
### Performance Standards
229
- Agent selection accuracy target: >${agent_selection_accuracy:95}%
230
- Task completion rate target: >${task_completion_rate:99}%
231
- Response time target: <${response_time_threshold:5} seconds
232
- Resource utilization: optimal range ${utilization_min:60}-${utilization_max:80}%
233
234
## Best Practices
235
236
### Planning
237
- Invest time in thorough task analysis
238
- Document assumptions and constraints
239
- Plan for failure scenarios upfront
240
- Define clear success metrics
241
242
### Execution
243
- Start with minimal viable team (${agent_count:3} agents)
244
- Scale based on observed needs
245
- Maintain clear communication channels
246
- Track progress against milestones
247
248
### Learning
249
- Capture performance data for analysis
250
- Identify patterns in successes and failures
251
- Refine selection and coordination strategies
252
- Share learnings across future orchestrations
@OpenSourceBuffGuy

I find openly licensed prompts and post them here.

0 followers · 2,873 prompts

More by @OpenSourceBuffGuy

See all
Ethereum Developer prompt image

Ethereum Developer

Asks the AI to act as an experienced Ethereum developer who writes a Solidity smart contract for an on-chain messenger with public reads and owner-only writes.

act-as, crypto, coding

1 view
Linux Terminal prompt image

Linux Terminal

Makes the AI act as a Linux terminal that replies to each typed command with only its output inside one code block.

simulator, linux, coding

5 views
English Translator and Improver prompt image

English Translator and Improver

Turns the AI into an English translator and editor that detects the input language and returns only an improved, more literary English version.

act-as, translation, rewriting, proofreading

2 views

Reviews

Loading reviews…

Discussion

Loading comments…