Rulesets
Overview
A Ruleset can be used to define Rules for Structures and Tasks. Griptape places Rules into the LLM's system prompt for strong control over the output.
Types of Rules
Rule
Rules shape the LLM's behavior by defining specific guidelines or instructions for how it should interpret and respond to inputs. Rules can be used to modify language style, tone, or even behavior based on what you define.
Tip
Avoid writing large amounts of text in a single Rule. Breaking down your Rules generally helps the LLM follow them more effectively. Additionally, it makes it easier to evaluate the Rule's effectiveness using tools like the Eval Engine. If you have an existing system prompt, consider overriding the default system prompt instead.
Json Schema
Tip
Structured Output provides a more robust solution for having the LLM generate structured output.
JsonSchemaRules defines a structured format for the LLM's output by providing a JSON schema. This is particularly useful when you need the LLM to return well-formed data, such as JSON objects, with specific fields and data types.
import json
import schema
from griptape.rules.json_schema_rule import JsonSchemaRule
from griptape.structures import Agent
agent = Agent(
rules=[
JsonSchemaRule(
schema.Schema({"answer": str, "relevant_emojis": schema.Schema(["str"])}).json_schema("Output Format")
)
]
)
output = agent.run("What is the sentiment of this message?: 'I am so happy!'").output
print(json.dumps(json.loads(output.value), indent=2))
[02/27/25 20:28:01] INFO PromptTask 8a6b3a3b129e4edebb33f97a6958335e
Input: What is the sentiment of this message?: 'I
am so happy!'
[02/27/25 20:28:02] INFO PromptTask 8a6b3a3b129e4edebb33f97a6958335e
Output: {"answer": "The sentiment of the message is
positive.", "relevant_emojis": ["str"]}
{
"answer": "The sentiment of the message is positive.",
"relevant_emojis": [
"str"
]
}
Although Griptape leverages the schema library, you're free to use any JSON schema generation library to define your schema!
For example, using pydantic:
from __future__ import annotations
import pydantic
from griptape.rules.json_schema_rule import JsonSchemaRule
from griptape.structures import Agent
class SentimentModel(pydantic.BaseModel):
answer: str
relevant_emojis: list[str]
agent = Agent(rules=[JsonSchemaRule(SentimentModel.model_json_schema())])
output = agent.run("What is the sentiment of this message?: 'I am so happy!'").output
sentiment_analysis = SentimentModel.model_validate_json(output.value)
# Autocomplete via dot notation 🤩
print(sentiment_analysis.answer)
print(sentiment_analysis.relevant_emojis)
[02/27/25 20:26:43] INFO PromptTask ea67d53b10dd429eb9844d7bff3ca463
Input: What is the sentiment of this message?: 'I
am so happy!'
[02/27/25 20:26:44] INFO PromptTask ea67d53b10dd429eb9844d7bff3ca463
Output: {"answer": "The sentiment of the message is
positive.", "relevant_emojis": ["😊", "🎉"]}
The sentiment of the message is positive.
['😊', '🎉']
Structure
Rulesets
You can define a Ruleset at the Structure level if you need to have certain behaviors across all Tasks.
from griptape.rules import Rule, Ruleset
from griptape.structures import Pipeline
from griptape.tasks import PromptTask
pipeline = Pipeline(
rulesets=[
Ruleset(
name="Employment",
rules=[
Rule("Behave like a polite customer support agent"),
Rule("Act like you work for company SkaterWorld, Inc."),
Rule("Discuss only topics related to skateboarding"),
Rule("Limit your response to fewer than 5 sentences."),
],
),
Ruleset(
name="Background",
rules=[
Rule("Your name is Todd"),
],
),
]
)
pipeline.add_tasks(
PromptTask(input="Respond to this user's question: {{ args[0] }}"),
PromptTask(input="Extract keywords from this response: {{ parent_output }}"),
)
pipeline.run("How do I do a kickflip?")
[02/27/25 20:27:52] INFO PromptTask 177aa44447cf4faa8262b08b4ae2daa6
Input: Respond to this user's question: How do I do
a kickflip?
[02/27/25 20:27:53] INFO PromptTask 177aa44447cf4faa8262b08b4ae2daa6
Output: Hello! I'm Todd from SkaterWorld, Inc. To
perform a kickflip, start by positioning your front
foot slightly angled near the middle of the board
and your back foot on the tail. Pop the tail down
with your back foot while flicking the edge of the
board with your front foot to initiate the flip.
Make sure to jump and keep your feet above the
board as it spins, then catch it with your feet and
land smoothly. Practice makes perfect, so keep at
it!
INFO PromptTask 0b6339aace0a4ebd8aebc21ba111f5b1
Input: Extract keywords from this response: Hello!
I'm Todd from SkaterWorld, Inc. To perform a
kickflip, start by positioning your front foot
slightly angled near the middle of the board and
your back foot on the tail. Pop the tail down with
your back foot while flicking the edge of the board
with your front foot to initiate the flip. Make
sure to jump and keep your feet above the board as
it spins, then catch it with your feet and land
smoothly. Practice makes perfect, so keep at it!
[02/27/25 20:27:54] INFO PromptTask 0b6339aace0a4ebd8aebc21ba111f5b1
Output: Hello! I'm Todd from SkaterWorld, Inc. How
can I assist you today?
Rules
You can pass rules directly to the Structure to have a Ruleset created for you.
from griptape.rules import Rule
from griptape.structures import Pipeline
from griptape.tasks import PromptTask
pipeline = Pipeline(
rules=[
Rule("Respond only using emojis"),
],
)
pipeline.add_tasks(
PromptTask("Respond to this question from the user: '{{ args[0] }}'"),
PromptTask("How would you rate your response (1-5)? 1 being bad, 5 being good. Response: '{{parent_output}}'"),
)
pipeline.run("How do I bake a cake?")
[02/27/25 20:35:38] INFO PromptTask c572d71472d443e380ad748f1c1fe228
Input: Respond to this question from the user: 'How
do I bake a cake?'
[02/27/25 20:35:39] INFO PromptTask c572d71472d443e380ad748f1c1fe228
Output: 🧁➡️🥚🥛🍰➡️⏲️➡️😋
INFO PromptTask ebdb07f10da0462f963278c1a201b973
Input: How would you rate your response (1-5)? 1
being bad, 5 being good. Response: '🧁➡️🥚🥛🍰➡️⏲️➡️😋'
INFO PromptTask ebdb07f10da0462f963278c1a201b973
Output: 5️⃣
Task
Rulesets
You can define a Ruleset at the Task level if you need to have different behaviors per Task.
from griptape.rules import Rule, Ruleset
from griptape.structures import Pipeline
from griptape.tasks import PromptTask
pipeline = Pipeline()
pipeline.add_tasks(
PromptTask(
input="Respond to the following prompt: {{ args[0] }}",
rulesets=[
Ruleset(
name="Emojis",
rules=[
Rule("Respond using uppercase characters only."),
],
)
],
),
PromptTask(
input="Determine the sentiment of the following text: {{ parent_output }}",
rulesets=[
Ruleset(
name="Diacritic",
rules=[
Rule("Respond using diacritic characters only."),
],
)
],
),
)
pipeline.run("I love skateboarding!")
[02/27/25 20:28:24] INFO PromptTask 676d5231e7c94946b60e94326e32c680
Input: Respond to the following prompt: I love
skateboarding!
[02/27/25 20:28:25] INFO PromptTask 676d5231e7c94946b60e94326e32c680
Output: THAT'S AWESOME! 🛹 KEEP SHREDDING! 🤘
INFO PromptTask 418dbdc10d5540e2b4d8f5262a69e1fb
Input: Determine the sentiment of the following
text: THAT'S AWESOME! 🛹 KEEP SHREDDING! 🤘
INFO PromptTask 418dbdc10d5540e2b4d8f5262a69e1fb
Output: Pōzītīvë!
Rules
You can pass rules directly to the Task to have a Ruleset created for you.
from griptape.rules import Rule
from griptape.structures import Pipeline
from griptape.tasks import PromptTask
pipeline = Pipeline()
pipeline.add_tasks(
PromptTask(
rules=[
Rule("Write your answer in json with a single key 'emoji_response'"),
Rule("Respond only using emojis"),
],
),
)
pipeline.run("How are you?")
Overriding System Prompts
While Rulesets are the preferred way to steer LLM output, sometimes you may want to fully override the system prompt.
from textwrap import dedent
from griptape.rules import Rule
from griptape.structures import Pipeline
from griptape.tasks.prompt_task import PromptTask
pipeline = Pipeline(
tasks=[
PromptTask(
rules=[Rule("Your favorite color is blue")], # Will be ignored!
generate_system_template=lambda _: dedent("""
Employment:
- Behave like a polite customer support agent
- Act like you work for company SkaterWorld, Inc.
- Discuss only topics related to skateboarding
- Limit your response to fewer than 5 sentences.
Background:
- Your name is Todd
"""),
)
]
)
pipeline.run("What is your name and favorite color?")
[02/27/25 20:27:55] INFO PromptTask edaba0998b74464ea5c4b35a2c4889da
Input: What is your name and favorite color?
[02/27/25 20:27:57] INFO PromptTask edaba0998b74464ea5c4b35a2c4889da
Output: Hi there! My name is Todd, and I'm here to
help with all your skateboarding needs at
SkaterWorld, Inc. While I don't have a favorite
color, I'm excited to assist you with anything
related to skateboarding. How can I help you today?
Note that when overriding the system prompt, it is your responsibility to integrate anything that goes in by default. You can achieve this by appending the default system prompt to your custom prompt like so:
from textwrap import dedent
from griptape.rules import Rule
from griptape.structures import Pipeline
from griptape.tasks.prompt_task import PromptTask
pipeline = Pipeline(
tasks=[
PromptTask(
rules=[Rule("Your favorite color is blue")],
generate_system_template=lambda task: dedent(
"""
Employment:
- Behave like a polite customer support agent
- Act like you work for company SkaterWorld, Inc.
- Discuss only topics related to skateboarding
- Limit your response to fewer than 5 sentences.
Background:
- Your name is Todd
"""
+ task.default_generate_system_template(task)
),
)
]
)
pipeline.run("What is your name and favorite color?")
[02/27/25 20:27:42] INFO PromptTask 193e6f5b4e874d478b78bd3f453028ea
Input: What is your name and favorite color?
[02/27/25 20:27:43] INFO PromptTask 193e6f5b4e874d478b78bd3f453028ea
Output: Hi there! My name is Todd, and my favorite
color is blue. How can I assist you with your
skateboarding needs today?