[Docs] Managing Agent Steps (#166)

langchain-ai · May 18, 2024 · c0e24d7 · c0e24d7
1 parent 012ce04
commit c0e24d7
Show file tree

Hide file tree

Showing 6 changed files with 675 additions and 42 deletions.
diff --git a/docs/docs/how-tos/index.md b/docs/docs/how-tos/index.md
@@ -2,10 +2,6 @@
 
 Welcome to the LangGraphJS How-to Guides! These guides provide practical, step-by-step instructions for accomplishing key tasks in LangGraphJS.
 
-## In progress
-
-🚧 This section is currently in progress. More updates to come! 🚧
-
 ## Core
 
 The core guides show how to address common needs when building a out AI workflows, with special focus placed on [ReAct](https://arxiv.org/abs/2210.03629)-style agents with [tool calling](https://js.langchain.com/v0.2/docs/how_to/tool_calling/).
@@ -28,3 +24,4 @@ The following examples are useful especially if you are used to LangChain's Agen
 - [Force calling a tool first](force-calling-a-tool-first.ipynb): Define a fixed workflow before ceding control to the ReAct agent
 - [Dynamic direct return](dynamically-returning-directly.ipynb): Let the LLM decide whether the graph should finish after a tool is run or whether the LLM should be able to review the output and keep going
 - [Respond in structured format](respond-in-format.ipynb): Let the LLM use tools or populate schema to provide the user. Useful if your agent should generate structured content
+- [Managing agent steps](managing-agent-steps.ipynb): How to format the intermediate steps of your workflow for the agent
diff --git a/docs/mkdocs.yml b/docs/mkdocs.yml
@@ -96,6 +96,7 @@ nav:
  - "how-tos/force-calling-a-tool-first.ipynb"
  - "how-tos/dynamically-returning-directly.ipynb"
  - "how-tos/respond-in-format.ipynb"
+ - "how-tos/managing-agent-steps.ipynb"
  - "Conceptual Guides":
  - "concepts/index.md"
  - "Reference":

diff --git a/examples/how-tos/dynamically-returning-directly.ipynb b/examples/how-tos/dynamically-returning-directly.ipynb
@@ -27,7 +27,7 @@
  "\n",
  "Next, we need to set API keys for OpenAI (the LLM we will use). Optionally, we\n",
  "can set API key for [LangSmith tracing](https://smith.langchain.com/), which\n",
- "will give us best-in-class observability."
+ "will give us best-in-class observability.\n"
  ]
  },
  {
@@ -60,12 +60,12 @@
  "that.\n",
  "\n",
  "To add a 'return_direct' option, we will create a custom zod schema to use\n",
- "**instead of** the schema that would be automatically inferred by the tool."
+ "**instead of** the schema that would be automatically inferred by the tool.\n"
  ]
  },
  {
  "cell_type": "code",
- "execution_count": 12,
+ "execution_count": 2,
  "id": "481c95ac",
  "metadata": {},
  "outputs": [],
@@ -111,12 +111,12 @@
  "We can now wrap these tools in a simple ToolExecutor.\\\n",
  "This is a real simple class that takes in a ToolInvocation and calls that tool,\n",
  "returning the output. A ToolInvocation is any type with `tool` and `toolInput`\n",
- "attribute."
+ "attribute.\n"
  ]
  },
  {
  "cell_type": "code",
- "execution_count": 13,
+ "execution_count": 3,
  "id": "250415e4",
  "metadata": {},
  "outputs": [],
@@ -142,12 +142,12 @@
  " [tool calling](https://js.langchain.com/v0.2/docs/concepts/#functiontool-calling).\n",
  "\n",
  "Note: these model requirements are not requirements for using LangGraph - they\n",
- "are just requirements for this one example."
+ "are just requirements for this one example.\n"
  ]
  },
  {
  "cell_type": "code",
- "execution_count": 14,
+ "execution_count": 4,
  "id": "2c24d018",
  "metadata": {},
  "outputs": [],
@@ -181,12 +181,12 @@
  "\n",
  "For this example, the state we will track will just be a list of messages. We\n",
  "want each node to just add messages to that list. Therefore, we will define the\n",
- "state as follows:"
+ "state as follows:\n"
  ]
  },
  {
  "cell_type": "code",
- "execution_count": 15,
+ "execution_count": 5,
  "id": "24454123",
  "metadata": {},
  "outputs": [],
@@ -233,12 +233,12 @@
  " agent to decide what to do next\n",
  "\n",
  "Let's define the nodes, as well as a function to decide how what conditional\n",
- "edge to take."
+ "edge to take.\n"
  ]
  },
  {
  "cell_type": "code",
- "execution_count": 16,
+ "execution_count": 6,
  "id": "23a8b9c6",
  "metadata": {},
  "outputs": [],
@@ -270,7 +270,7 @@
  " const response = await boundModel.invoke(messages, config);\n",
  " // We return an object, because this will get added to the existing list\n",
  " return { messages: [response] };\n",
- "};"
+ "};\n"
  ]
  },
  {
@@ -285,7 +285,7 @@
  },
  {
  "cell_type": "code",
- "execution_count": 17,
+ "execution_count": 7,
  "id": "05203811",
  "metadata": {},
  "outputs": [
@@ -388,7 +388,7 @@
  "}"
  ]
  },
- "execution_count": 17,
+ "execution_count": 7,
  "metadata": {},
  "output_type": "execute_result"
  }
@@ -436,12 +436,12 @@
  "\n",
  "We can now use it! This now exposes the\n",
  "[same interface](https://js.langchain.com/docs/expression_language/) as all\n",
- "other LangChain runnables."
+ "other LangChain runnables.\n"
  ]
  },
  {
  "cell_type": "code",
- "execution_count": 18,
+ "execution_count": 8,
  "id": "de5f4864",
  "metadata": {},
  "outputs": [
@@ -451,7 +451,13 @@
  "text": [
  "[human]: what is the weather in sf\n",
  "-----\n",
- "\n",
+ "\n"
+ ]
+ },
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
  "[ai]: [object Object] \n",
  "Tools: \n",
  "- search({\"query\":\"weather in sf\",\"return_direct\":true})\n",
@@ -464,13 +470,17 @@
  }
  ],
  "source": [
- "import { AIMessage, BaseMessage, HumanMessage } from \"@langchain/core/messages\";\n",
+ "import {\n",
+ " AIMessage,\n",
+ " BaseMessage,\n",
+ " HumanMessage,\n",
+ " isAIMessage,\n",
+ "} from \"@langchain/core/messages\";\n",
  "\n",
  "const prettyPrint = (message: BaseMessage) => {\n",
  " let txt = `[${message._getType()}]: ${message.content}`;\n",
  " if (\n",
- " (message._getType() === \"ai\" &&\n",
- " (message as AIMessage)?.tool_calls?.length) ||\n",
+ " (isAIMessage(message) && (message as AIMessage)?.tool_calls?.length) ||\n",
  " 0 > 0\n",
  " ) {\n",
  " const tool_calls = (message as AIMessage)?.tool_calls\n",
@@ -486,12 +496,12 @@
  " const lastMessage = output.messages[output.messages.length - 1];\n",
  " prettyPrint(lastMessage);\n",
  " console.log(\"-----\\n\");\n",
- "}\n"
+ "}"
  ]
  },
  {
  "cell_type": "code",
- "execution_count": 19,
+ "execution_count": 9,
  "id": "986f8cfe",
  "metadata": {},
  "outputs": [
@@ -535,7 +545,7 @@
  "id": "51fa73e6",
  "metadata": {},
  "source": [
- "Done! The graph **stopped** after running the `tools` node!"
+ "Done! The graph **stopped** after running the `tools` node!\n"
  ]
  }
  ],

diff --git a/examples/how-tos/human-in-the-loop.ipynb b/examples/how-tos/human-in-the-loop.ipynb
@@ -392,12 +392,17 @@
  }
  ],
  "source": [
- "import { AIMessage, BaseMessage, HumanMessage } from \"@langchain/core/messages\";\n",
+ "import {\n",
+ " AIMessage,\n",
+ " BaseMessage,\n",
+ " HumanMessage,\n",
+ " isAIMessage,\n",
+ "} from \"@langchain/core/messages\";\n",
  "\n",
  "const prettyPrint = (message: BaseMessage) => {\n",
  " let txt = `[${message._getType()}]: ${message.content}`;\n",
  " if (\n",
- " message._getType() === \"ai\" && (message as AIMessage)?.tool_calls?.length ||\n",
+ " isAIMessage(message) && (message as AIMessage)?.tool_calls?.length ||\n",
  " 0 > 0\n",
  " ) {\n",
  " const tool_calls = (message as AIMessage)?.tool_calls?.map(\n",