Python API

banks.prompt.Prompt

The Prompt class is the only thing you need to know about Banks on the Python side. The class can be initialized with a string variable containing the prompt template text, then you can invoke the method text on your instance to pass the data needed to render the template and get back the final prompt.

A quick example:

from banks import Prompt


p = Prompt("Write a 500-word blog post on {{ topic }}.")
my_topic = "retrogame computing"
print(p.text({"topic": my_topic}))

raw `property`

raw: str

Returns the raw text of the prompt.

init

__init__(text: str, *, name: str | None = None, version: str | None = None, metadata: dict[str, Any] | None = None, canary_word: str | None = None, render_cache: RenderCache | None = None) -> None

Prompt constructor.

Parameters:

Name	Type	Description	Default
`text`	`str`	The template text.	required
`name`	`str \| None`	The name to identify this prompt.	`None`
`version`	`str \| None`	The version string attached to this prompt.	`None`
`metadata`	`dict[str, Any] \| None`	A key-value set of metadata pairs attached to this prompt.	`None`
`canary_word`	`str \| None`	The string to use for the `{{canary_word}}` extension. If `None`, a default string will be generated.	`None`
`render_cache`	`RenderCache \| None`	The caching backend to store rendered prompts. If `None`, the default in-memory backend will be used.	`None`

canary_leaked

canary_leaked(text: str) -> bool

Returns whether the canary word is present in text, signalling the prompt might have leaked.

chat_messages

chat_messages(data: dict[str, Any] | None = None) -> list[ChatMessage]

Render the prompt using variables present in data

Parameters:

Name	Type	Description	Default
`data`	`dict[str, Any] \| None`	A dictionary containing the context variables.	`None`

text

text(data: dict[str, Any] | None = None) -> str

Render the prompt using variables present in data

Parameters:

Name	Type	Description	Default
`data`	`dict[str, Any] \| None`	A dictionary containing the context variables.	`None`

banks.prompt.AsyncPrompt

Banks provides async support through the machinery provided by Jinja

Since the Jinja environment is a global state in banks, the library can work either with or without async support, and this must be known before importing anything.

If the application using banks runs within an asyncio loop, you can do two things to optimize banks' execution:

Set the environment variable BANKS_ASYNC_ENABLED=true.
Use the AsyncPrompt class that has an awaitable run method.

For example, let's render a prompt that contains some calls to the generate extension. Those calls will be heavily I/O bound, so other tasks can take advantage and advance while the prompt is being rendered.

Example:

# Enable async support before importing from banks
import os

os.environ["BANKS_ASYNC_ENABLED"] = "true"

# Show logs to see what's happening at runtime
import logging

logging.basicConfig(level=logging.INFO)

import asyncio
from banks import AsyncPrompt

prompt_template = """
Generate a tweet about the topic '{{ topic }}' with a positive sentiment.
"""

async def task(task_id: int, sleep_time: int):
    logging.info(f"Task {task_id} is running.")
    await asyncio.sleep(sleep_time)
    logging.info(f"Task {task_id} done.")


async def main():
    p = AsyncPrompt(prompt_template)
    # Schedule the prompt rendering along with two executions of 'task', one sleeping for 10 seconds
    # and one sleeping for 1 second
    results = await asyncio.gather(p.text({"topic": "AI frameworks"}), task(1, 10), task(2, 1))
    print("All tasks done, rendered prompt:")
    print(results[0])


asyncio.run(main())

text `async`

text(data: dict[str, Any] | None = None) -> str

Render the prompt using variables present in data

Parameters:

Name	Type	Description	Default
`data`	`dict[str, Any] \| None`	A dictionary containing the context variables.	`None`

banks.registries.directory.DirectoryPromptRegistry

Registry that stores prompts as files in a directory structure.

path `property`

path: Path

Get the directory path where prompts are stored.

init

__init__(directory_path: Path, *, force_reindex: bool = False)

Initialize the directory prompt registry.

Parameters:

Name	Type	Description	Default
`directory_path`	`Path`	Path to directory where prompts will be stored	required
`force_reindex`	`bool`	Whether to force rebuilding the index from disk	`False`

Raises:

Type	Description
`ValueError`	If directory_path is not a directory

get

get(*, name: str, version: str | None = None) -> Prompt

Retrieve a prompt by name and version.

Parameters:

Name	Type	Description	Default
`name`	`str`	Name of the prompt to retrieve	required
`version`	`str \| None`	Version of the prompt (optional)	`None`

Returns:

Type	Description
`Prompt`	The requested Prompt object

Raises:

Type	Description
`PromptNotFoundError`	If prompt doesn't exist

set

set(*, prompt: Prompt, overwrite: bool = False)

Store a prompt in the registry.

Parameters:

Name	Type	Description	Default
`prompt`	`Prompt`	The Prompt object to store	required
`overwrite`	`bool`	Whether to overwrite existing prompt	`False`

Raises:

Type	Description
`InvalidPromptError`	If prompt exists and overwrite=False

banks.registries.file.FilePromptRegistry

A prompt registry storing all prompt data in a single JSON file.

init

__init__(registry_index: Path) -> None

Initialize the file prompt registry.

Parameters:

Name	Type	Description	Default
`registry_index`	`Path`	Path to the JSON file that will store the prompts	required

Note

Creates parent directories if they don't exist.

get

get(*, name: str, version: str | None = None) -> Prompt

Retrieve a prompt by name and version.

Parameters:

Name	Type	Description	Default
`name`	`str`	Name of the prompt to retrieve	required
`version`	`str \| None`	Version of the prompt (optional)	`None`

Returns:

Type	Description
`Prompt`	The requested Prompt object

Raises:

Type	Description
`PromptNotFoundError`	If the requested prompt doesn't exist

set

set(*, prompt: Prompt, overwrite: bool = False) -> None

Store a prompt in the registry.

Parameters:

Name	Type	Description	Default
`prompt`	`Prompt`	The Prompt object to store	required
`overwrite`	`bool`	Whether to overwrite an existing prompt	`False`

Raises:

Type	Description
`InvalidPromptError`	If prompt exists and overwrite=False

banks.registries.redis.RedisPromptRegistry

A prompt registry that stores prompts in Redis.

init

__init__(redis_url: str = 'redis://localhost:6379', prefix: str = 'banks:prompt:') -> None

Initialize the Redis prompt registry.

Parameters:

Name	Type	Description	Default
`redis_url`	`str`	Redis connection URL	`'redis://localhost:6379'`
`prefix`	`str`	Key prefix for storing prompts in Redis	`'banks:prompt:'`

get

get(*, name: str, version: str | None = None) -> Prompt

Get a prompt by name and version.

Parameters:

Name	Type	Description	Default
`name`	`str`	Name of the prompt	required
`version`	`str \| None`	Version of the prompt (optional)	`None`

Returns:

Type	Description
`Prompt`	Prompt instance

Raises:

Type	Description
`PromptNotFoundError`	If prompt doesn't exist

set

set(*, prompt: Prompt, overwrite: bool = False) -> None

Store a prompt in Redis.

Parameters:

Name	Type	Description	Default
`prompt`	`Prompt`	Prompt instance to store	required
`overwrite`	`bool`	Whether to overwrite existing prompt	`False`

Raises:

Type	Description
`InvalidPromptError`	If prompt exists and overwrite=False

Python API

banks.prompt.Prompt

raw property

__init__

canary_leaked

chat_messages

text

banks.prompt.AsyncPrompt

text async

banks.registries.directory.DirectoryPromptRegistry

path property

__init__

get

set

banks.registries.file.FilePromptRegistry

__init__

get

set

banks.registries.redis.RedisPromptRegistry

__init__

get

set

raw `property`

init

text `async`

path `property`

init

init

init