Tasks
Overview
A Task is a purpose-built abstraction for the Large Language Model (LLM). Griptape offers various types of Tasks, each suitable for specific use cases.
Context
Tasks that take input have a field input which lets you define the Task objective. Within the input, you can access the following context variables:
args: an array of arguments passed to the.run()method.structure: the structure that the task belongs to.- user defined context variables
Additional context variables may be added based on the Structure running the task.
from griptape.structures import Agent
from griptape.tasks import PromptTask
agent = Agent()
agent.add_task(
PromptTask(
"Respond to the user's following question '{{ args[0] }}' in the language '{{preferred_language}}' and tone '{{tone}}'.",
context={"preferred_language": "ENGLISH", "tone": "PLAYFUL"},
)
)
agent.run("How do I bake a cake?")
[09/08/23 11:12:47] INFO PromptTask 0f5a5def49864126834627b6140f3e63
Input: Respond to the user's following question 'How do I bake a cake?' in the language 'ENGLISH' and tone
'PLAYFUL'.
[09/08/23 11:13:17] INFO PromptTask 0f5a5def49864126834627b6140f3e63
Output: Oh, you're in for a treat! Baking a cake is like creating a masterpiece, but way more delicious! Here's a
simple recipe to get you started:
1. Preheat your oven to 350°F (175°C). It's like sunbathing, but for your cake!
2. Grab a bowl and mix together 2 cups of sugar and 1/2 cup of softened butter. It's like making sweet, buttery
sandcastles!
3. Crack in 3 eggs, one at a time, and stir in 2 teaspoons of vanilla extract. It's a pool party in your bowl!
4. In a separate bowl, combine 1 1/2 cups of all-purpose flour, 1 3/4 teaspoons of baking powder, and a pinch of
salt. This is the dry gang!
5. Gradually mix the dry gang into the buttery pool party. Stir until it's just combined, we don't want to
overwork the partygoers!
6. Pour the batter into a greased cake pan. It's like tucking your cake into bed!
7. Bake for 30 to 40 minutes, or until a toothpick comes out clean. It's like playing hide and seek with your
cake!
8. Let it cool, then frost and decorate as you like. This is where you can let your creativity shine!
Remember, baking is all about having fun and enjoying the process. So, put on your favorite tunes, roll up your
sleeves, and let's get baking! 🍰🎉
Prompt Task
For general purpose prompting, use the PromptTask:
from griptape.tasks import PromptTask
from griptape.structures import Agent
agent = Agent()
agent.add_task(
# take the first argument from the agent `run` method
PromptTask("Respond to the users following request: {{ args[0] }}"),
)
agent.run("Write me a haiku")
[10/20/23 15:27:26] INFO PromptTask f5025c6352914e9f80ef730e5269985a
Input: Respond to the users following request:
Write me a haiku
[10/20/23 15:27:28] INFO PromptTask f5025c6352914e9f80ef730e5269985a
Output: Gentle morning dew,
Kisses the waking flowers,
Day begins anew.
Toolkit Task
To use Griptape Tools, use a Toolkit Task. This Task takes in one or more Tools which the LLM will decide to use through Chain of Thought (CoT) reasoning. Because this Task uses CoT, it is recommended to only use with very capable models.
from griptape.tasks import ToolkitTask
from griptape.structures import Agent
from griptape.tools import WebScraper, FileManager, TaskMemoryClient
agent = Agent()
agent.add_task(
ToolkitTask(
"Load https://www.griptape.ai, summarize it, and store it in a file called griptape.txt",
tools=[WebScraper(), FileManager(), TaskMemoryClient(off_prompt=False)]
),
)
agent.run()
[09/08/23 11:14:55] INFO ToolkitTask 22af656c6ad643e188fe80f9378dfff9
Input: Load https://www.griptape.ai, summarize it, and store it in a file called griptape.txt
[09/08/23 11:15:02] INFO Subtask 7a6356470e6a4b08b61edc5591b37f0c
Thought: The first step is to load the webpage using the WebScraper tool's get_content activity.
Action: {"name": "WebScraper", "path": "get_content", "input": {"values": {"url":
"https://www.griptape.ai"}}}
[09/08/23 11:15:03] INFO Subtask 7a6356470e6a4b08b61edc5591b37f0c
Response: Output of "WebScraper.get_content" was stored in memory with memory_name "TaskMemory" and
artifact_namespace "2b50373849d140f698ba8071066437ee"
[09/08/23 11:15:11] INFO Subtask a22a7e4ebf594b4b895fcbe8a95c1dd3
Thought: Now that the webpage content is stored in memory, I can use the TaskMemory tool's summarize activity
to summarize it.
Action: {"name": "TaskMemoryClient", "path": "summarize", "input": {"values": {"memory_name": "TaskMemory", "artifact_namespace": "2b50373849d140f698ba8071066437ee"}}}
[09/08/23 11:15:15] INFO Subtask a22a7e4ebf594b4b895fcbe8a95c1dd3
Response: Griptape is an open source framework that allows developers to build and deploy AI applications
using large language models (LLMs). It provides the ability to create conversational and event-driven apps that
can access and manipulate data securely. Griptape enforces structures like sequential pipelines and DAG-based
workflows for predictability, while also allowing for creativity by safely prompting LLMs with external APIs and
data stores. The framework can be used to create AI systems that operate across both dimensions. Griptape Cloud
is a managed platform for deploying and managing AI apps, and it offers features like scheduling and connecting
to data stores and APIs.
[09/08/23 11:15:27] INFO Subtask 7afb3d44d0114b7f8ef2dac4314a8e90
Thought: Now that I have the summary, I can use the FileManager tool's save_file_to_disk activity to store the
summary in a file named griptape.txt.
Action: {"name": "FileManager", "path": "save_file_to_disk", "input": {"values":
{"memory_name": "TaskMemory", "artifact_namespace": "2b50373849d140f698ba8071066437ee", "path":
"griptape.txt"}}}
INFO Subtask 7afb3d44d0114b7f8ef2dac4314a8e90
Response: saved successfully
[09/08/23 11:15:31] INFO ToolkitTask 22af656c6ad643e188fe80f9378dfff9
Output: The summary of the webpage https://www.griptape.ai has been successfully stored in a file named
griptape.txt.
Tool Task
Another way to use Griptape Tools, is with a Tool Task. This Task takes in a single Tool which the LLM will use without Chain of Thought (CoT) reasoning. Because this Task does not use CoT, it is better suited for less capable models.
from griptape.structures import Agent
from griptape.tasks import ToolTask
from griptape.tools import Calculator
# Initialize the agent and add a task
agent = Agent()
agent.add_task(ToolTask(tool=Calculator()))
# Run the agent with a prompt
agent.run("Give me the answer for 5*4.")
[10/20/23 14:20:25] INFO ToolTask df1604b417a84ee781dbd1f2b904ed30
Input: Give me the answer for 5*4.
[10/20/23 14:20:29] INFO Subtask a9a9ad7be2bf465fa82bd350116fabe4
Action: {
"name": "Calculator",
"path": "calculate",
"input": {
"values": {
"expression": "5*4"
}
}
}
[10/20/23 14:20:30] INFO Subtask a9a9ad7be2bf465fa82bd350116fabe4
Response: 20
INFO ToolTask df1604b417a84ee781dbd1f2b904ed30
Output: 20
Extraction Task
To extract information from text, use an ExtractionTask. This Task takes an Extraction Engine, and a set of arguments specific to the Engine.
CSV Extraction
from griptape.drivers import OpenAiChatPromptDriver
from griptape.tasks import ExtractionTask
from griptape.structures import Agent
from griptape.engines import CsvExtractionEngine
# Instantiate the CSV extraction engine
csv_extraction_engine = CsvExtractionEngine(
prompt_driver=OpenAiChatPromptDriver(model="gpt-3.5-turbo")
)
# Define some unstructured data and columns
csv_data = """
Alice, 28, lives in New York.
Bob, 35 lives in California.
Charlie is 40 and lives in Texas.
"""
columns = ["Name", "Age", "Address"]
# Create an agent and add the ExtractionTask to it
agent = Agent()
agent.add_task(
ExtractionTask(
extraction_engine=csv_extraction_engine,
args={"column_names": columns},
)
)
# Run the agent
agent.run(csv_data)
[12/19/23 10:33:11] INFO ExtractionTask e87fb457edf8423ab8a78583badd7a11
Input:
Alice, 28, lives in New York.
Bob, 35 lives in California.
Charlie is 40 and lives in Texas.
[12/19/23 10:33:13] INFO ExtractionTask e87fb457edf8423ab8a78583badd7a11
Output: Name,Age,Address
Alice,28,New York
Bob,35,California
Charlie,40,Texas
JSON Extraction
from schema import Schema
from griptape.drivers import OpenAiChatPromptDriver
from griptape.tasks import ExtractionTask
from griptape.structures import Agent
from griptape.engines import JsonExtractionEngine
# Instantiate the json extraction engine
json_extraction_engine = JsonExtractionEngine(
prompt_driver=OpenAiChatPromptDriver(model="gpt-3.5-turbo"),
)
# Define some unstructured data and a schema
json_data = """
Alice (Age 28) lives in New York.
Bob (Age 35) lives in California.
"""
user_schema = Schema(
{"users": [{"name": str, "age": int, "location": str}]}
).json_schema("UserSchema")
agent = Agent()
agent.add_task(
ExtractionTask(
extraction_engine=json_extraction_engine,
args={"template_schema": user_schema},
)
)
# Run the agent
agent.run(json_data)
[12/19/23 10:37:41] INFO ExtractionTask 3315cc77f94943a2a2dceccfe44f6a67
Input:
Alice (Age 28) lives in New York.
Bob (Age 35) lives in California.
[12/19/23 10:37:44] INFO ExtractionTask 3315cc77f94943a2a2dceccfe44f6a67
Output: {'name': 'Alice', 'age': 28, 'location': 'New York'}
{'name': 'Bob', 'age': 35, 'location': 'California'}
Text Summary Task
To summarize a text, use the TextSummaryTask. This Task takes an Summarization Engine, and a set of arguments to the engine.
from griptape.structures import Agent
from griptape.tasks import TextSummaryTask
# Create a new agent
agent = Agent()
# Add the TextSummaryTask to the agent
agent.add_task(TextSummaryTask())
# Run the agent
agent.run(
"Artificial Intelligence (AI) is a branch of computer science that deals with "
"creating machines capable of thinking and learning. It encompasses various fields "
"such as machine learning, neural networks, and deep learning. AI has the potential "
"to revolutionize many sectors, including healthcare, finance, and transportation. "
"Our life in this modern age depends largely on computers. It is almost impossible "
"to think about life without computers. We need computers in everything that we use "
"in our daily lives. So it becomes very important to make computers intelligent so "
"that our lives become easy. Artificial Intelligence is the theory and development "
"of computers, which imitates the human intelligence and senses, such as visual "
"perception, speech recognition, decision-making, and translation between languages."
" Artificial Intelligence has brought a revolution in the world of technology. "
)
[10/20/23 15:37:46] INFO TextSummaryTask e870f2a6226f43fcb89f93b1c0c85b10
Input: Artificial Intelligence (AI) is a branch of
computer science that deals with creating machines
capable of thinking and learning. It encompasses
various fields such as machine learning, neural
networks, and deep learning. AI has the potential
to revolutionize many sectors, including
healthcare, finance, and transportation. Our life
in this modern age depends largely on computers. It
is almost impossible to think about life without
computers. We need computers in everything that we
use in our daily lives. So it becomes very
important to make computers intelligent so that our
lives become easy. Artificial Intelligence is the
theory and development of computers, which imitates
the human intelligence and senses, such as visual
perception, speech recognition, decision-making,
and translation between languages. Artificial
Intelligence has brought a revolution in the world
of technology.
[10/20/23 15:37:49] INFO TextSummaryTask e870f2a6226f43fcb89f93b1c0c85b10
Output: Artificial Intelligence (AI) is a branch of
computer science that focuses on creating
intelligent machines. It encompasses various fields
such as machine learning and neural networks. AI
has the potential to revolutionize sectors like
healthcare, finance, and transportation. It is
essential to make computers intelligent to simplify
our daily lives. AI imitates human intelligence and
senses, bringing a revolution in technology.
Text Query Task
To query text, use the TextQueryTask. This Task takes a Query Engine, and a set of arguments specific to the engine.
from griptape.drivers import OpenAiChatPromptDriver
from griptape.structures import Agent
from griptape.tasks import TextQueryTask
from griptape.drivers import LocalVectorStoreDriver, OpenAiEmbeddingDriver
from griptape.engines import VectorQueryEngine
from griptape.artifacts import TextArtifact
# Initialize Embedding Driver and Vector Store Driver
vector_store_driver = LocalVectorStoreDriver(embedding_driver=OpenAiEmbeddingDriver())
artifact = TextArtifact(
"Griptape builds AI-powered applications that connect securely to your enterprise data and APIs."
"Griptape Agents provide incredible power and flexibility when working with large language models."
)
# Create a VectorQueryEngine using the LocalVectorStoreDriver
vector_query_engine = VectorQueryEngine(
vector_store_driver=vector_store_driver,
prompt_driver=OpenAiChatPromptDriver(model="gpt-3.5-turbo")
)
vector_query_engine.upsert_text_artifact(artifact=artifact)
# Instantiate the agent and add TextQueryTask with the VectorQueryEngine
agent = Agent()
agent.add_task(
TextQueryTask(
"Respond to the users following query: {{ args[0] }}",
query_engine=vector_query_engine,
)
)
# Run the agent with a query string
agent.run("Give me information about Griptape")
[10/20/23 15:32:39] INFO TextQueryTask a1d2eceab9204679b3f701f6ea821606
Input: Respond to the users following query: Give
me information about Griptape
[10/20/23 15:32:41] INFO TextQueryTask a1d2eceab9204679b3f701f6ea821606
Output: Griptape builds AI-powered applications
that connect securely to your enterprise data and
APIs. Griptape Agents provide incredible power and
flexibility when working with large language
models.
Code Execution Task
To execute an arbitrary Python function, use the CodeExecutionTask. This task takes a python function, and authors can elect to return a custom artifact.
from griptape.structures import Pipeline
from griptape.tasks import CodeExecutionTask, PromptTask
from griptape.artifacts import BaseArtifact, TextArtifact
def character_counter(task: CodeExecutionTask) -> BaseArtifact:
result = len(task.input)
# For functions that don't need to return anything, we recommend returning task.input
return TextArtifact(str(result))
# Instantiate the pipeline
pipeline = Pipeline()
pipeline.add_tasks(
# take the first argument from the pipeline `run` method
CodeExecutionTask(run_fn=character_counter),
# # take the output from the previous task and insert it into the prompt
PromptTask("{{args[0]}} using {{ parent_output }} characters")
)
pipeline.run("Write me a line in a poem")
[01/09/24 15:23:54] INFO CodeExecutionTask 048b1f548683475187064dde90055f72
Input: Write me a line in a poem
INFO CodeExecutionTask 048b1f548683475187064dde90055f72
Output: 25
INFO PromptTask b6156dc5c0c6404488ab925989e78b01
Input: Write me a line in a poem using 25
characters
[01/09/24 15:24:03] INFO PromptTask b6156dc5c0c6404488ab925989e78b01
Output: "Silent code, loud impact."
Image Generation Tasks
To generate an image, use one of the following Image Generation Tasks. All Image Generation Tasks accept an Image Generation Engine configured to use an Image Generation Driver.
All successful Image Generation Tasks will always output an Image Artifact. Each task can be configured to additionally write the generated image to disk by providing either the output_file or output_dir field. The output_file field supports file names in the current directory (my_image.png), relative directory prefixes (images/my_image.png), or absolute paths (/usr/var/my_image.png). By setting output_dir, the task will generate a file name and place the image in the requested directory.
Prompt Image Generation Task
The Prompt Image Generation Task generates an image from a text prompt.
from griptape.engines import PromptImageGenerationEngine
from griptape.drivers import OpenAiImageGenerationDriver
from griptape.tasks import PromptImageGenerationTask
from griptape.structures import Pipeline
# Create a driver configured to use OpenAI's DALL-E 3 model.
driver = OpenAiImageGenerationDriver(
model="dall-e-3",
quality="hd",
style="natural",
)
# Create an engine configured to use the driver.
engine = PromptImageGenerationEngine(
image_generation_driver=driver,
)
# Instantiate a pipeline.
pipeline = Pipeline()
# Add a PromptImageGenerationTask to the pipeline.
pipeline.add_tasks(
PromptImageGenerationTask(
input="{{ args[0] }}",
image_generation_engine=engine,
output_dir="images/",
)
)
pipeline.run("An image of a mountain on a summer day")
Variation Image Generation Task
The Variation Image Generation Task generates an image using an input image and a text prompt. The input image is used as a basis for generating a new image as requested by the text prompt.
from griptape.engines import VariationImageGenerationEngine
from griptape.drivers import AmazonBedrockImageGenerationDriver, \
BedrockStableDiffusionImageGenerationModelDriver
from griptape.tasks import VariationImageGenerationTask
from griptape.loaders import ImageLoader
from griptape.structures import Pipeline
# Create a driver configured to use Stable Diffusion via Bedrock.
driver = AmazonBedrockImageGenerationDriver(
image_generation_model_driver=BedrockStableDiffusionImageGenerationModelDriver(),
model="stability.stable-diffusion-xl-v0",
)
# Create an engine configured to use the driver.
engine = VariationImageGenerationEngine(
image_generation_driver=driver,
)
# Load input image artifact.
with open("tests/resources/mountain.png", "rb") as f:
image_artifact = ImageLoader().load(f.read())
# Instatiate a pipeline.
pipeline = Pipeline()
# Add a VariationImageGenerationTask to the pipeline.
pipeline.add_task(
VariationImageGenerationTask(
input=("{{ args[0] }}", image_artifact),
image_generation_engine=engine,
output_dir="images/",
)
)
pipeline.run("An image of a mountain landscape on a snowy winter day")
Inpainting Image Generation Task
The Inpainting Image Generation Task generates an image using an input image, a mask image, and a text prompt. The input image will be modified within the bounds of the mask image as requested by the text prompt.
from griptape.engines import InpaintingImageGenerationEngine
from griptape.drivers import AmazonBedrockImageGenerationDriver, \
BedrockStableDiffusionImageGenerationModelDriver
from griptape.tasks import InpaintingImageGenerationTask
from griptape.loaders import ImageLoader
from griptape.structures import Pipeline
# Create a driver configured to use Stable Diffusion via Bedrock.
driver = AmazonBedrockImageGenerationDriver(
image_generation_model_driver=BedrockStableDiffusionImageGenerationModelDriver(),
model="stability.stable-diffusion-xl-v0",
)
# Create an engine configured to use the driver.
engine = InpaintingImageGenerationEngine(
image_generation_driver=driver,
)
# Load input image artifacts.
with open("tests/resources/mountain.png", "rb") as f:
image_artifact = ImageLoader().load(f.read())
with open("tests/resources/mountain-mask.png", "rb") as f:
mask_artifact = ImageLoader().load(f.read())
# Instantiate a pipeline.
pipeline = Pipeline()
# Add an InpaintingImageGenerationTask to the pipeline.
pipeline.add_task(
InpaintingImageGenerationTask(
input=("{{ args[0] }}", image_artifact, mask_artifact),
image_generation_engine=engine,
output_dir="images/"
)
)
pipeline.run("An image of a castle built into the side of a mountain")
Outpainting Image Generation Task
The Outpainting Image Generation Task generates an image using an input image, a mask image, and a text prompt. The input image will be modified outside the bounds of a mask image as requested by the text prompt.
from griptape.engines import OutpaintingImageGenerationEngine
from griptape.drivers import AmazonBedrockImageGenerationDriver, \
BedrockStableDiffusionImageGenerationModelDriver
from griptape.tasks import OutpaintingImageGenerationTask
from griptape.loaders import ImageLoader
from griptape.structures import Pipeline
# Create a driver configured to use Stable Diffusion via Bedrock.
driver = AmazonBedrockImageGenerationDriver(
image_generation_model_driver=BedrockStableDiffusionImageGenerationModelDriver(),
model="stability.stable-diffusion-xl-v0",
)
# Create an engine configured to use the driver.
engine = OutpaintingImageGenerationEngine(
image_generation_driver=driver,
)
# Load input image artifacts.
with open("tests/resources/mountain.png", "rb") as f:
image_artifact = ImageLoader().load(f.read())
with open("tests/resources/mountain-mask.png", "rb") as f:
mask_artifact = ImageLoader().load(f.read())
# Instantiate a pipeline.
pipeline = Pipeline()
# Add an OutpaintingImageGenerationTask to the pipeline.
pipeline.add_task(
OutpaintingImageGenerationTask(
input=("{{ args[0] }}", image_artifact, mask_artifact),
image_generation_engine=engine,
output_dir="images/",
)
)
pipeline.run("An image of a mountain shrouded by clouds")
Image Query Task
The Image Query Task executes a natural language query on one or more input images. This Task uses an Image Query Engine configured with an Image Query Driver to perform the query. The functionality provided by this Task depend on the capabilities of the model provided by the Driver.
This Task accepts two inputs: a query (represented by either a string or a Text Artifact) and a list of Image Artifacts or a Callable returning these two values.
from griptape.engines import ImageQueryEngine
from griptape.drivers import OpenAiVisionImageQueryDriver
from griptape.tasks import ImageQueryTask
from griptape.loaders import ImageLoader
from griptape.structures import Pipeline
# Create a driver configured to use OpenAI's GPT-4 Vision model.
driver = OpenAiVisionImageQueryDriver(
model="gpt-4-vision-preview",
max_tokens=100,
)
# Create an engine configured to use the driver.
engine = ImageQueryEngine(
image_query_driver=driver,
)
# Load the input image artifact.
with open("tests/resources/mountain.png", "rb") as f:
image_artifact = ImageLoader().load(f.read())
# Instantiate a pipeline.
pipeline = Pipeline()
# Add an ImageQueryTask to the pipeline.
pipeline.add_task(
ImageQueryTask(
input=("{{ args[0] }}", [image_artifact]),
image_query_engine=engine,
)
)
pipeline.run("Describe the weather in the image")
Structure Run Task
The Structure Run Task executes another Structure with a given input. This Task is useful for orchestrating multiple specialized Structures in a single run. Note that the input to the Task is a tuple of arguments that will be passed to the Structure.
import os
from griptape.rules import Rule, Ruleset
from griptape.structures import Agent, Pipeline
from griptape.tasks import StructureRunTask
from griptape.drivers import LocalStructureRunDriver
from griptape.tools import (
TaskMemoryClient,
WebScraper,
WebSearch,
)
def build_researcher():
researcher = Agent(
tools=[
WebSearch(
google_api_key=os.environ["GOOGLE_API_KEY"],
google_api_search_id=os.environ["GOOGLE_API_SEARCH_ID"],
off_prompt=False,
),
WebScraper(
off_prompt=True,
),
TaskMemoryClient(off_prompt=False),
],
rulesets=[
Ruleset(
name="Position",
rules=[
Rule(
value="Senior Research Analyst",
)
],
),
Ruleset(
name="Objective",
rules=[
Rule(
value="Uncover cutting-edge developments in AI and data science",
)
],
),
Ruleset(
name="Background",
rules=[
Rule(
value="""You work at a leading tech think tank.,
Your expertise lies in identifying emerging trends.
You have a knack for dissecting complex data and presenting actionable insights."""
)
],
),
Ruleset(
name="Desired Outcome",
rules=[
Rule(
value="Full analysis report in bullet points",
)
],
),
],
)
return researcher
def build_writer():
writer = Agent(
input_template="Instructions: {{args[0]}}\nContext: {{args[1]}}",
rulesets=[
Ruleset(
name="Position",
rules=[
Rule(
value="Tech Content Strategist",
)
],
),
Ruleset(
name="Objective",
rules=[
Rule(
value="Craft compelling content on tech advancements",
)
],
),
Ruleset(
name="Backstory",
rules=[
Rule(
value="""You are a renowned Content Strategist, known for your insightful and engaging articles.
You transform complex concepts into compelling narratives."""
)
],
),
Ruleset(
name="Desired Outcome",
rules=[
Rule(
value="Full blog post of at least 4 paragraphs",
)
],
),
],
)
return writer
team = Pipeline(
tasks=[
StructureRunTask(
(
"""Perform a detailed examination of the newest developments in AI as of 2024.
Pinpoint major trends, breakthroughs, and their implications for various industries.""",
),
driver=LocalStructureRunDriver(structure_factory_fn=build_researcher),
),
StructureRunTask(
(
"""Utilize the gathered insights to craft a captivating blog
article showcasing the key AI innovations.
Ensure the content is engaging yet straightforward, appealing to a tech-aware readership.
Keep the tone appealing and use simple language to make it less technical.""",
"{{parent_output}}",
),
driver=LocalStructureRunDriver(structure_factory_fn=build_writer),
),
],
)
team.run()