Deprecate ToolkitTask, add functionality to PromptTask (#1463)

griptape-ai · Dec 20, 2024 · 23ac354 · 23ac354
1 parent d8ace6f
commit 23ac354
Show file tree

Hide file tree

Showing 44 changed files with 421 additions and 409 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -22,6 +22,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Rulesets can now be serialized and deserialized.
 - `ToolkitTask` now serializes its `tools` field.
 - `PromptTask.prompt_driver` is now serialized.
+- `PromptTask` can now do everything a `ToolkitTask` can do.
 
 ### Fixed
 
@@ -32,6 +33,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Parsing `ActionCallDeltaMessageContent`s with empty string `partial_input`s.
 - `Stream` util not properly propagating thread contextvars.
 
+### Deprecated
+
+- `ToolkitTask`. `PromptTask` is a drop-in replacement.
+
 ## [1.0.0] - 2024-12-09
 
 ### Added

diff --git a/README.md b/README.md
@@ -111,7 +111,7 @@ agent.run("https://griptape.ai", "griptape.txt")
 And here is the output:
 
 ```
-[08/12/24 14:48:15] INFO     ToolkitTask c90d263ec69046e8b30323c131ae4ba0
+[08/12/24 14:48:15] INFO     PromptTask c90d263ec69046e8b30323c131ae4ba0
                              Input: Load https://griptape.ai, summarize it, and store it in a file called griptape.txt.
 [08/12/24 14:48:16] INFO     Subtask ebe23832cbe2464fb9ecde9fcee7c30f
                              Actions: [
@@ -166,7 +166,7 @@ And here is the output:
                              ]
                     INFO     Subtask c233853450fb4fd6a3e9c04c52b33bf6
                              Response: Successfully saved memory artifacts to disk
-[08/12/24 14:48:23] INFO     ToolkitTask c90d263ec69046e8b30323c131ae4ba0
+[08/12/24 14:48:23] INFO     PromptTask c90d263ec69046e8b30323c131ae4ba0
                              Output: The content from https://griptape.ai has been summarized and stored in a file called `griptape.txt`.
 ```
 
@@ -177,8 +177,8 @@ The important thing to note here is that no matter how big the webpage is it can
 In the above example, we set [off_prompt](https://docs.griptape.ai/stable/griptape-framework/structures/task-memory.md#off-prompt) to `True`, which means that the LLM can never see the data it manipulates, but can send it to other Tools.
 
 > [!IMPORTANT]
-> This example uses Griptape's [ToolkitTask](https://docs.griptape.ai/stable/griptape-framework/structures/tasks/#toolkit-task), which requires a highly capable LLM to function correctly. By default, Griptape uses the [OpenAiChatPromptDriver](https://docs.griptape.ai/stable/griptape-framework/drivers/prompt-drivers/#openai-chat); for another powerful LLM try swapping to the [AnthropicPromptDriver](https://docs.griptape.ai/stable/griptape-framework/drivers/prompt-drivers/#anthropic)!
-> If you're using a less powerful LLM, consider using the [ToolTask](https://docs.griptape.ai/stable/griptape-framework/structures/tasks/#tool-task) instead, as the `ToolkitTask` might not work properly or at all.
+> This example uses Griptape's [PromptTask](https://docs.griptape.ai/stable/griptape-framework/structures/tasks/#prompt-task) with `tools`, which requires a highly capable LLM to function correctly. By default, Griptape uses the [OpenAiChatPromptDriver](https://docs.griptape.ai/stable/griptape-framework/drivers/prompt-drivers/#openai-chat); for another powerful LLM try swapping to the [AnthropicPromptDriver](https://docs.griptape.ai/stable/griptape-framework/drivers/prompt-drivers/#anthropic)!
+> If you're using a less powerful LLM, consider using the [ToolTask](https://docs.griptape.ai/stable/griptape-framework/structures/tasks/#tool-task) instead, as the `PromptTask` with `tools` might not work properly or at all.
 
 [Check out our docs](https://docs.griptape.ai/stable/griptape-framework/drivers/prompt-drivers/) to learn more about how to use Griptape with other LLM providers like Anthropic, Claude, Hugging Face, and Azure.
 

diff --git a/docs/griptape-framework/drivers/web-search-drivers.md b/docs/griptape-framework/drivers/web-search-drivers.md
@@ -16,7 +16,7 @@ You can use Web Search Drivers with [Structures](../structures/agents.md):
 ```
 
 ```
-ToolkitTask 45a53f1024494baab41a1f10a67017b1
+PromptTask 45a53f1024494baab41a1f10a67017b1
     Output: Here are some websites with information about AI
     frameworks:
 

diff --git a/docs/griptape-framework/index.md b/docs/griptape-framework/index.md
@@ -106,7 +106,7 @@ Agents on their own are fun, but let's add some capabilities to them using Gript
 Here is the chain of thought from the Agent. Notice where it realizes it can use the tool you just injected to do the calculation.[^1]
 
 ```
-[07/23/24 10:47:38] INFO     ToolkitTask 6a51060d1fb74e57840a91aa319f26dc
+[07/23/24 10:47:38] INFO     PromptTask 6a51060d1fb74e57840a91aa319f26dc
                              Input: what is 7^12
 [07/23/24 10:47:39] INFO     Subtask 0c984616fd2345a7b48a0b0d692daa3c
                              Actions: [
@@ -123,7 +123,7 @@ Here is the chain of thought from the Agent. Notice where it realizes it can use
                              ]
                     INFO     Subtask 0c984616fd2345a7b48a0b0d692daa3c
                              Response: 13841287201
-[07/23/24 10:47:40] INFO     ToolkitTask 6a51060d1fb74e57840a91aa319f26dc
+[07/23/24 10:47:40] INFO     PromptTask 6a51060d1fb74e57840a91aa319f26dc
                              Output: 13,841,287,201
 Answer: 13,841,287,201
 ```
@@ -137,7 +137,7 @@ Agents are great for getting started, but they are intentionally limited to a si
 ```
 
 ```
-[08/12/24 14:50:28] INFO     ToolkitTask 19dcf6020968468a91aa8a93c2a3f645
+[08/12/24 14:50:28] INFO     PromptTask 19dcf6020968468a91aa8a93c2a3f645
                              Input: Load https://www.griptape.ai, summarize it, and store it in griptape.txt
 [08/12/24 14:50:30] INFO     Subtask a685799379c5421b91768353fc219939
                              Actions: [
@@ -202,7 +202,7 @@ Agents are great for getting started, but they are intentionally limited to a si
                              ]
                     INFO     Subtask aaaeca1a089844d4915d065deb3c00cf
                              Response: Successfully saved file
-[08/12/24 14:50:39] INFO     ToolkitTask 19dcf6020968468a91aa8a93c2a3f645
+[08/12/24 14:50:39] INFO     PromptTask 19dcf6020968468a91aa8a93c2a3f645
                              Output: The content from https://www.griptape.ai has been summarized and stored in griptape.txt.
                     INFO     PromptTask dbbb38f144f445db896dc12854f17ad3
                              Input: Say the following in spanish: The content from https://www.griptape.ai has been summarized and stored in griptape.txt.

diff --git a/docs/griptape-framework/misc/src/events_3.py b/docs/griptape-framework/misc/src/events_3.py
@@ -1,7 +1,7 @@
 from griptape.drivers import OpenAiChatPromptDriver
 from griptape.events import BaseChunkEvent, EventBus, EventListener
 from griptape.structures import Pipeline
-from griptape.tasks import ToolkitTask
+from griptape.tasks import PromptTask
 from griptape.tools import PromptSummaryTool, WebScraperTool
 
 EventBus.add_event_listeners(
@@ -15,7 +15,7 @@
 
 pipeline = Pipeline()
 pipeline.add_tasks(
-    ToolkitTask(
+    PromptTask(
         "Based on https://griptape.ai, tell me what griptape is.",
         prompt_driver=OpenAiChatPromptDriver(model="gpt-4o", stream=True),
         tools=[WebScraperTool(off_prompt=True), PromptSummaryTool(off_prompt=False)],

diff --git a/docs/griptape-framework/misc/src/events_chunk_stream.py b/docs/griptape-framework/misc/src/events_chunk_stream.py
@@ -1,7 +1,7 @@
 from griptape.drivers import OpenAiChatPromptDriver
 from griptape.events import ActionChunkEvent, EventBus, EventListener, TextChunkEvent
 from griptape.structures import Pipeline
-from griptape.tasks import ToolkitTask
+from griptape.tasks import PromptTask
 from griptape.tools import PromptSummaryTool, WebScraperTool
 
 EventBus.add_event_listeners(
@@ -19,7 +19,7 @@
 
 pipeline = Pipeline()
 pipeline.add_tasks(
-    ToolkitTask(
+    PromptTask(
         "Based on https://griptape.ai, tell me what griptape is.",
         prompt_driver=OpenAiChatPromptDriver(model="gpt-4o", stream=True),
         tools=[WebScraperTool(off_prompt=True), PromptSummaryTool(off_prompt=False)],

diff --git a/docs/griptape-framework/src/index_4.py b/docs/griptape-framework/src/index_4.py
@@ -1,14 +1,14 @@
 from griptape.memory.structure import ConversationMemory
 from griptape.structures import Pipeline
-from griptape.tasks import PromptTask, ToolkitTask
+from griptape.tasks import PromptTask
 from griptape.tools import FileManagerTool, PromptSummaryTool, WebScraperTool
 
 # Pipelines represent sequences of tasks.
 pipeline = Pipeline(conversation_memory=ConversationMemory())
 
 pipeline.add_tasks(
     # Load up the first argument from `pipeline.run`.
-    ToolkitTask(
+    PromptTask(
         "{{ args[0] }}",
         # Add tools for web scraping, and file management
         tools=[WebScraperTool(off_prompt=True), FileManagerTool(off_prompt=True), PromptSummaryTool(off_prompt=False)],

diff --git a/docs/griptape-framework/structures/agents.md b/docs/griptape-framework/structures/agents.md
@@ -7,21 +7,33 @@ search:
 
 An [Agent](../../reference/griptape/structures/agent.md) is the quickest way to get started with Griptape.
 Agents take in [tools](../../reference/griptape/structures/agent.md#griptape.structures.agent.Agent.tools) and [input](../../reference/griptape/structures/agent.md#griptape.structures.agent.Agent.input)
-directly, which the agent uses to dynamically determine whether to use a [Prompt Task](./tasks.md#prompt-task) or [Toolkit Task](./tasks.md#toolkit-task).
-
-If [tools](../../reference/griptape/structures/agent.md#griptape.structures.agent.Agent.tools) are passed provided to the Agent, a [Toolkit Task](./tasks.md#toolkit-task) will be used. If no [tools](../../reference/griptape/structures/agent.md#griptape.structures.agent.Agent.tools)
-are provided, a [Prompt Task](./tasks.md#prompt-task) will be used.
+directly, which the agent uses to add a [Prompt Task](./tasks.md#prompt-task).
 
 You can access the final output of the Agent by using the [output](../../reference/griptape/structures/structure.md#griptape.structures.structure.Structure.output) attribute.
 
-## Toolkit Task Agent
+### Agent Input
+
+```python
+--8<-- "docs/griptape-framework/structures/src/agents_2.py"
+```
+
+```
+[09/08/23 10:10:24] INFO     PromptTask e70fb08090b24b91a9307fa83479e851
+                             Input: Write me a haiku about Skateboards and Programming
+[09/08/23 10:10:28] INFO     PromptTask e70fb08090b24b91a9307fa83479e851
+                             Output: Code on wheels in flight,
+                             Skateboards meet algorithms bright,
+                             In binary, we ignite.
+```
+
+### Agent Tools
 
 ```python
 --8<-- "docs/griptape-framework/structures/src/agents_1.py"
 ```
 
 ```
-[07/23/24 10:53:41] INFO     ToolkitTask 487db777bc014193ba90b061451b69a6
+[07/23/24 10:53:41] INFO     PromptTask 487db777bc014193ba90b061451b69a6
                              Input: Calculate the following: what's 13^7?
 [07/23/24 10:53:42] INFO     Subtask 126cefa3ac5347b88495e25af52f3268
                              Actions: [
@@ -38,22 +50,7 @@ You can access the final output of the Agent by using the [output](../../referen
                              ]
                     INFO     Subtask 126cefa3ac5347b88495e25af52f3268
                              Response: 62748517
-[07/23/24 10:53:43] INFO     ToolkitTask 487db777bc014193ba90b061451b69a6
+[07/23/24 10:53:43] INFO     PromptTask 487db777bc014193ba90b061451b69a6
                              Output: 62,748,517
 Answer: 62,748,517
 ```
-
-## Prompt Task Agent
-
-```python
---8<-- "docs/griptape-framework/structures/src/agents_2.py"
-```
-
-```
-[09/08/23 10:10:24] INFO     PromptTask e70fb08090b24b91a9307fa83479e851
-                             Input: Write me a haiku about Skateboards and Programming
-[09/08/23 10:10:28] INFO     PromptTask e70fb08090b24b91a9307fa83479e851
-                             Output: Code on wheels in flight,
-                             Skateboards meet algorithms bright,
-                             In binary, we ignite.
-```
diff --git a/docs/griptape-framework/structures/src/agents_1.py b/docs/griptape-framework/structures/src/agents_1.py
@@ -1,7 +1,7 @@
 from griptape.structures import Agent
-from griptape.tools import CalculatorTool
 
-agent = Agent(input="Calculate the following: {{ args[0] }}", tools=[CalculatorTool()])
+agent = Agent(
+    input="Write me a {{ args[0] }} about {{ args[1] }} and {{ args[2] }}",
+)
 
-agent.run("what's 13^7?")
-print("Answer:", agent.output)
+agent.run("Haiku", "Skateboards", "Programming")
diff --git a/docs/griptape-framework/structures/src/agents_2.py b/docs/griptape-framework/structures/src/agents_2.py
@@ -1,11 +1,7 @@
 from griptape.structures import Agent
-from griptape.tasks import PromptTask
+from griptape.tools import CalculatorTool
 
-agent = Agent()
-agent.add_task(
-    PromptTask(
-        "Write me a {{ creative_medium }} about {{ args[0] }} and {{ args[1] }}", context={"creative_medium": "haiku"}
-    )
-)
+agent = Agent(input="Calculate the following: {{ args[0] }}", tools=[CalculatorTool()])
 
-agent.run("Skateboards", "Programming")
+agent.run("what's 13^7?")
+print("Answer:", agent.output)
diff --git a/docs/griptape-framework/structures/src/tasks_4.py b/docs/griptape-framework/structures/src/tasks_4.py
@@ -1,10 +1,10 @@
 from griptape.structures import Agent
-from griptape.tasks import ToolkitTask
+from griptape.tasks import PromptTask
 from griptape.tools import FileManagerTool, PromptSummaryTool, WebScraperTool
 
 agent = Agent()
 agent.add_task(
-    ToolkitTask(
+    PromptTask(
         "Load https://www.griptape.ai, summarize it, and store it in a file called griptape.txt",
         tools=[WebScraperTool(off_prompt=True), FileManagerTool(off_prompt=True), PromptSummaryTool(off_prompt=True)],
     ),

diff --git a/docs/griptape-framework/structures/task-memory.md b/docs/griptape-framework/structures/task-memory.md
@@ -26,14 +26,14 @@ Lets look at a simple example where `off_prompt` is set to `False`:
 ```
 
 ```
-[04/26/24 13:06:42] INFO     ToolkitTask 36b9dea13b9d479fb752014f41dca54c
+[04/26/24 13:06:42] INFO     PromptTask 36b9dea13b9d479fb752014f41dca54c
                              Input: What is the square root of 12345?
 [04/26/24 13:06:48] INFO     Subtask a88c0feeaef6493796a9148ed68c9caf
                              Thought: To find the square root of 12345, I can use the CalculatorTool action with the expression "12345 ** 0.5".
                              Actions: [{"name": "CalculatorTool", "path": "calculate", "input": {"values": {"expression": "12345 ** 0.5"}}, "tag": "sqrt_12345"}]
                     INFO     Subtask a88c0feeaef6493796a9148ed68c9caf
                              Response: 111.1080555135405
-[04/26/24 13:06:49] INFO     ToolkitTask 36b9dea13b9d479fb752014f41dca54c
+[04/26/24 13:06:49] INFO     PromptTask 36b9dea13b9d479fb752014f41dca54c
                              Output: The square root of 12345 is approximately 111.108.
 ```
 
@@ -46,7 +46,7 @@ Let's explore what happens when `off_prompt` is set to `True`:
 ```
 
 ```
-[04/26/24 13:07:02] INFO     ToolkitTask ecbb788d9830491ab72a8a2bbef5fb0a
+[04/26/24 13:07:02] INFO     PromptTask ecbb788d9830491ab72a8a2bbef5fb0a
                              Input: What is the square root of 12345?
 [04/26/24 13:07:10] INFO     Subtask 4700dc0c2e934d1a9af60a28bd770bc6
                              Thought: To find the square root of a number, we can use the CalculatorTool action with the expression "sqrt(12345)". However, the CalculatorTool
@@ -85,7 +85,7 @@ If we had kept it as `True`, the results would have been stored back Task Memory
 ```
 
 ```
-[08/12/24 14:54:04] INFO     ToolkitTask f7ebd8acc3d64e3ca9db82ef9ec4e65f
+[08/12/24 14:54:04] INFO     PromptTask f7ebd8acc3d64e3ca9db82ef9ec4e65f
                              Input: What is the square root of 12345?
 [08/12/24 14:54:05] INFO     Subtask 777693d039e74ed288f663742fdde2ea
                              Actions: [
@@ -121,7 +121,7 @@ If we had kept it as `True`, the results would have been stored back Task Memory
                              ]
 [08/12/24 14:54:07] INFO     Subtask c8394ca51f1f4ae1b715618a2c5c8120
                              Response: The text contains a single numerical value: 111.1080555135405.
-[08/12/24 14:54:08] INFO     ToolkitTask f7ebd8acc3d64e3ca9db82ef9ec4e65f
+[08/12/24 14:54:08] INFO     PromptTask f7ebd8acc3d64e3ca9db82ef9ec4e65f
                              Output: The square root of 12345 is approximately 111.108.
 ```
 
@@ -139,7 +139,7 @@ Let's say we want to query the contents of a very large webpage.
 When running this example, we get the following error:
 
 ```
-[04/26/24 13:20:02] ERROR    ToolkitTask 67e2f907f95d4850ae79f9da67df54c1
+[04/26/24 13:20:02] ERROR    PromptTask 67e2f907f95d4850ae79f9da67df54c1
                              Error code: 400 - {'error': {'message': "This model's maximum context length is 8192 tokens. However, your messages resulted in 73874 tokens.
                              Please reduce the length of the messages.", 'type': 'invalid_request_error', 'param': 'messages', 'code': 'context_length_exceeded'}}
 ```
@@ -154,7 +154,7 @@ Note that we're setting `off_prompt` to `False` on the `QueryTool` so that the _
 And now we get the expected output:
 
 ```
-[08/12/24 14:56:18] INFO     ToolkitTask d3ce58587dc944b0a30a205631b82944
+[08/12/24 14:56:18] INFO     PromptTask d3ce58587dc944b0a30a205631b82944
                              Input: According to this page https://en.wikipedia.org/wiki/Elden_Ring, how many copies of Elden Ring have been sold?
 [08/12/24 14:56:20] INFO     Subtask 494850ec40fe474c83d48b5620c5dcbb
                              Actions: [
@@ -192,7 +192,7 @@ And now we get the expected output:
 [08/12/24 14:56:29] INFO     Subtask 8669ee523bb64550850566011bcd14e2
                              Response: "Elden Ring" sold 13.4 million copies worldwide by the end of March 2022 and 25 million by June 2024. The downloadable content (DLC)
                              "Shadow of the Erdtree" sold five million copies within three days of its release.
-[08/12/24 14:56:30] INFO     ToolkitTask d3ce58587dc944b0a30a205631b82944
+[08/12/24 14:56:30] INFO     PromptTask d3ce58587dc944b0a30a205631b82944
                              Output: Elden Ring sold 13.4 million copies worldwide by the end of March 2022 and 25 million by June 2024.
 ```
 
@@ -208,7 +208,7 @@ In this example, GPT-4 _never_ sees the contents of the page, only that it was s
 ```
 
 ```
-[08/12/24 14:55:21] INFO     ToolkitTask 329b1abc760e4d30bbf23e349451d930
+[08/12/24 14:55:21] INFO     PromptTask 329b1abc760e4d30bbf23e349451d930
                              Input: Use this page https://en.wikipedia.org/wiki/Elden_Ring to find how many copies of Elden Ring have been sold, and then save the result to
                              a file.
 [08/12/24 14:55:23] INFO     Subtask 26205b5623174424b618abafd886c4d8
@@ -265,7 +265,7 @@ In this example, GPT-4 _never_ sees the contents of the page, only that it was s
                              ]
                     INFO     Subtask 7aafcb3fb0d845858e2fcf9b8dc8a7ec
                              Response: Successfully saved memory artifacts to disk
-[08/12/24 14:55:40] INFO     ToolkitTask 329b1abc760e4d30bbf23e349451d930
+[08/12/24 14:55:40] INFO     PromptTask 329b1abc760e4d30bbf23e349451d930
                              Output: Successfully saved the number of copies sold of Elden Ring to a file named "elden_ring_sales.txt" in the "results" directory.
 ```