> ## Documentation Index
> Fetch the complete documentation index at: https://praison.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Simple Sequential Team (YAML)
> The copy-paste starting point for a multi-agent team that runs tasks in order
Two agents run in order: a researcher gathers facts, then a writer summarizes them.
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
graph LR
subgraph "Sequential Team"
R[π€ Researcher\nresearch_task] --> W[π€ Writer\nsummary_task]
W --> OUT[β
Summary]
end
classDef agent fill:#8B0000,stroke:#7C90A0,color:#fff
classDef output fill:#10B981,stroke:#7C90A0,color:#fff
class R,W agent
class OUT output
```
## Quick Start
Save this as `agents.yaml` in an empty folder. Tasks run in the order they are declared β the researcher's task first, then the writer's.
```yaml theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
# yaml-language-server: $schema=https://docs.praison.ai/schemas/agents.schema.json
# Minimal sequential team: researcher -> writer
# Run: praisonai agents.yaml
framework: "praisonai"
topic: "renewable energy"
roles:
researcher:
role: "Senior Researcher"
goal: "Find accurate information about {topic}"
backstory: "Expert at concise, factual research."
tasks:
research_task:
description: "Research {topic} and list 5 key facts."
expected_output: "A bullet list of 5 facts."
writer:
role: "Report Writer"
goal: "Turn research into readable summaries about {topic}"
backstory: "Clear technical writer."
tasks:
summary_task:
description: "Write a 3-sentence executive summary from the research."
expected_output: "3 sentences, plain text."
# Ordering is by task declaration order above (researcher's task runs first,
# then the writer's), which is how the roles-file loader sequences tasks.
# The optional `dependencies:` block below documents intent; the roles-file
# loader does not consume it, so keep tasks in the order you want them run.
dependencies:
- task: summary_task
depends_on: research_task
```
Set your API key and run the file with the `praisonai` CLI:
```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
export OPENAI_API_KEY=sk-...
praisonai agents.yaml
```
Change the subject by editing the `topic:` line β it is interpolated into every `{topic}` placeholder.
***
## How It Works
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
sequenceDiagram
participant User
participant CLI as praisonai CLI
participant Researcher
participant Writer
User->>CLI: praisonai agents.yaml
CLI->>Researcher: research_task (runs first)
Researcher-->>CLI: 5 facts
CLI->>Writer: summary_task (runs next)
Writer-->>CLI: 3-sentence summary
CLI-->>User: final output
```
| Fact | Detail |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Where tasks live | `roles..tasks.` β each role owns its tasks |
| Ordering | By **declaration order**, not the optional top-level `dependencies:` block. The roles-file loader (`AgentsGenerator.generate_crew_and_kickoff` in `praisonai/agents_generator.py`) does **not** consume `dependencies:`. |
| `{topic}` interpolation | The top-level `topic:` value is substituted into every `{topic}` in role, goal, and task fields. |
The single-brace `{topic}` interpolation here belongs to the **roles-file YAML loader**. The Python `AgentTeam(variables={...})` path uses **double-brace** `{{key}}` placeholders instead β don't mix the two.
***
## Python equivalent
The same team using friendly top-level imports:
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import Agent, Task, PraisonAIAgents
researcher = Agent(role="Senior Researcher", goal="Find accurate information")
writer = Agent(role="Report Writer", goal="Turn research into summaries")
t1 = Task(description="Research renewable energy and list 5 key facts", agent=researcher)
t2 = Task(description="Write a 3-sentence summary from the research", agent=writer, context=[t1])
PraisonAIAgents(agents=[researcher, writer], tasks=[t1, t2]).start()
```
***
## When to use this vs a workflow YAML
| You need⦠| Use | Example |
| ------------------------------------- | ---------------------------------------- | ------------------------------------------------------- |
| 2β5 agents, sequential / hierarchical | this recipe (`roles:` + nested `tasks:`) | this page |
| Routing by classifier output | workflow YAML | [Workflow Routing](/docs/features/workflow-routing) |
| Parallel branches | workflow YAML | [Workflow Parallelization](/docs/features/workflow-parallel) |
| Loops / planning / memory | workflow YAML | [YAML Workflows](/docs/features/yaml-workflows) |
***
## Related
Loops, planning, memory, and routing in YAML
Chain task outputs into the next step
Configure iteration and retry limits
Add lifecycle callbacks between tasks