Single Agent

Learn how to create a basic single-purpose AI agent for simple tasks.

The simplest possible Agent — one instruction, no tools, one call. This is the plain Agent showcase.

python

from praisonaiagents import Agent

agent = Agent(
    name="ContentWriter",
    instructions="You are a content writer. Output in markdown format.",
)

agent.start("Write a short blog post about AI assistants")

Diagram

Single-purpose agent for content generation. Minimal setup, no external tools.

Quick Start

Simple Usage

Create an agent with one instruction and call it.

python

from praisonaiagents import Agent

agent = Agent(
    name="ContentWriter",
    instructions="You are a content writer. Output in markdown format.",
)

agent.start("Write a short blog post about AI assistants")

With Configuration

Add memory and structured output for richer behaviour.

python

from praisonaiagents import Agent, Task, AgentTeam
from pydantic import BaseModel

class BlogPost(BaseModel):
    title: str
    content: str
    tags: list[str]

agent = Agent(
    name="ContentWriter",
    instructions="Write structured blog posts.",
    memory=True,
)

task = Task(
    description="Write a short blog post about AI assistants",
    expected_output="Structured blog post",
    agent=agent,
    output_pydantic=BlogPost,
)

AgentTeam(agents=[agent], tasks=[task]).start()

How It Works

Diagram

Simple

Agents: 1 — Single task requires only one agent.

Workflow

Receive input prompt
Process with LLM
Return generated content

Setup

bash

pip install praisonaiagents praisonai
export OPENAI_API_KEY="your-key"

Run — Python

python

from praisonaiagents import Agent

agent = Agent(
    name="ContentWriter",
    instructions="You are a content writer. Output in markdown format."
)

result = agent.start("Write a short blog post about AI assistants")
print(result)

Run — CLI

bash

praisonai "Write a short blog post about AI assistants"

Run — agents.yaml

yaml

framework: praisonai
topic: Content Generation
roles:
  content_writer:
    role: Content Writer
    goal: Generate engaging content
    backstory: You are an expert content writer
    tasks:
      write_content:
        description: Write a short blog post about AI assistants
        expected_output: A markdown formatted blog post

bash

praisonai agents.yaml

Serve API

python

from praisonaiagents import Agent

agent = Agent(
    name="ContentWriter",
    instructions="You are a content writer. Output in markdown format."
)

agent.launch(port=8080)

bash

curl -X POST http://localhost:8080/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "Write a haiku about coding"}'

Advanced Workflow (All Features)

Agents: 1 — Single agent with memory, persistence, structured output, and session resumability.

Workflow

Initialize session with unique ID for resumability
Configure SQLite persistence for conversation history
Process input with structured Pydantic output
Store results in memory for future context
Resume session later with same session_id

Setup

bash

pip install praisonaiagents praisonai pydantic
export OPENAI_API_KEY="your-key"

Run — Python

python

from praisonaiagents import Agent, Task, AgentTeam, Session
from pydantic import BaseModel

# Structured output schema
class BlogPost(BaseModel):
    title: str
    content: str
    tags: list[str]

# Create session for resumability
session = Session(session_id="blog-session-001", user_id="user-1")

# Agent with memory enabled
agent = Agent(
    name="ContentWriter",
    instructions="You are a content writer. Output structured JSON.",
    memory=True
)

# Task with structured output
task = Task(
    description="Write a short blog post about AI assistants",
    expected_output="Structured blog post",
    agent=agent,
    output_pydantic=BlogPost
)

# Run with SQLite persistence
agents = AgentTeam(
    agents=[agent],
    tasks=[task],
    memory=True
)

result = agents.start()
print(result)

# Resume later with same session_id
session2 = Session(session_id="blog-session-001", user_id="user-1")
context = session2.get_context()

Run — CLI

bash

# With memory and verbose
praisonai "Write a blog post about AI" --memory --verbose

# With session persistence
praisonai "Continue the blog post" --memory --session blog-session-001

Run — agents.yaml

yaml

framework: praisonai
topic: Content Generation
memory: true
memory_config:
  provider: sqlite
  db_path: content.db
roles:
  content_writer:
    role: Content Writer
    goal: Generate engaging content with structured output
    backstory: You are an expert content writer
    memory: true
    tasks:
      write_content:
        description: Write a short blog post about AI assistants
        expected_output: Structured blog post with title, content, and tags
        output_json:
          title: string
          content: string
          tags: array

bash

praisonai agents.yaml --verbose

Serve API

python

from praisonaiagents import Agent

agent = Agent(
    name="ContentWriter",
    instructions="You are a content writer.",
    memory=True
)

# Launch with persistence
agent.launch(port=8080)

bash

curl -X POST http://localhost:8080/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "Write a blog post", "session_id": "blog-001"}'

Save Output to File

Save agent responses to files using different methods:

write_file Tool

Agent decides when to save:

python

from praisonaiagents import Agent
from praisonaiagents.tools import write_file

agent = Agent(
    name="Writer",
    instructions="Write content and save to files",
    tools=[write_file]
)
agent.start("Write a poem and save it to poem.txt")

Task output_file

Auto-save task result:

python

from praisonaiagents import Agent, Task, AgentTeam

agent = Agent(name="Writer")
task = Task(
    description="Write a poem",
    agent=agent,
    output_file="poem.txt",
    create_directory=True
)
agents = AgentTeam(agents=[agent], tasks=[task])
agents.start()

Manual

Full control:

python

agent = Agent(name="Writer")
response = agent.start("Write a poem")
with open("poem.txt", "w") as f:
    f.write(response)

Monitor / Verify

bash

# Verbose output
praisonai "test prompt" --verbose

# Check telemetry
praisonai "test prompt" --telemetry

Cleanup

bash

rm -f content.db

Features Demonstrated

Feature	Implementation
Workflow	Single-step content generation
DB Persistence	SQLite via `memory_config`
Observability	`--verbose` flag
Resumability	`Session` with `session_id`
Structured Output	Pydantic `BlogPost` model

Best Practices

Start here before reaching for tools

A plain Agent handles most text tasks. Add tools only when the job genuinely needs live data or actions the model cannot perform alone.

Keep instructions specific

The single instruction shapes every response. A precise role — “content writer, output Markdown” — beats a vague “be helpful” for consistent results.

Add memory for multi-turn work

Set memory=True when a task spans several messages so the agent keeps context instead of treating each call as new.

Save output when generating at scale

Use a Task with output_file or a file-writing tool so generated content lands on disk automatically.

Card

Add a search tool for real-time information.

Card

Chain prompts for multi-step workflows.

Single Agent

Quick Start

How It Works

Simple

Workflow

Setup

Run — Python

Run — CLI

Run — agents.yaml

Serve API

Advanced Workflow (All Features)

Workflow

Setup

Run — Python

Run — CLI

Run — agents.yaml

Serve API

Save Output to File

write_file Tool

Task output_file

Manual

Monitor / Verify

Cleanup

Features Demonstrated

Best Practices

Related