github/MemGPT
7
0
Fork 0
mirror of https://github.com/cpacker/MemGPT.git synced 2025-06-03 04:30:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs: update examples in docs (#2396)

Browse Source
This commit is contained in:
cthomas 2025-01-29 14:48:14 -08:00 committed by GitHub
parent 9eab81570b
commit 7583cd0ffa
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
18 changed files with 343 additions and 222 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
.composio.lock Normal file
View File

@ -0,0 +1 @@
{}

134
examples/Building agents with Letta.ipynb
View File

@ -19,7 +19,7 @@
"metadata": {},
"source": [
"## Setup a Letta client \n",
"Make sure you run `pip install letta` and `letta quickstart`"
"Make sure you run `pip install letta_client` and start letta server `letta quickstart`"
]
},
{
@ -29,8 +29,9 @@
"metadata": {},
"outputs": [],
"source": [
"!pip install letta_client\n",
"!pip install letta\n",
"! letta quickstart"
"!letta quickstart"
]
},
{
@ -40,22 +41,9 @@
"metadata": {},
"outputs": [],
"source": [
"from letta import create_client \n",
"from letta_client import CreateBlock, Letta, MessageCreate \n",
"\n",
"client = create_client() "
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "9a28e38a-7dbe-4530-8260-202322a8458e",
"metadata": {},
"outputs": [],
"source": [
"from letta import LLMConfig, EmbeddingConfig\n",
"\n",
"client.set_default_llm_config(LLMConfig.default_config(\"gpt-4o-mini\")) \n",
"client.set_default_embedding_config(EmbeddingConfig.default_config(provider=\"openai\")) "
"client = Letta(base_url=\"http://localhost:8283\")"
]
},
{
@ -92,14 +80,20 @@
"metadata": {},
"outputs": [],
"source": [
"from letta.schemas.memory import ChatMemory\n",
"\n",
"agent_state = client.create_agent(\n",
"agent_state = client.agents.create(\n",
" name=agent_name, \n",
" memory=ChatMemory(\n",
" human=\"My name is Sarah\", \n",
" persona=\"You are a helpful assistant that loves emojis\"\n",
" )\n",
" memory_blocks=[\n",
" CreateBlock(\n",
" label=\"human\",\n",
" value=\"My name is Sarah\",\n",
" ),\n",
" CreateBlock(\n",
" label=\"persona\",\n",
" value=\"You are a helpful assistant that loves emojis\",\n",
" ),\n",
" ]\n",
" model=\"openai/gpt-4o-mini\",\n",
" embedding=\"openai/text-embedding-ada-002\",\n",
")"
]
},
@ -110,10 +104,14 @@
"metadata": {},
"outputs": [],
"source": [
"response = client.send_message(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" message=\"hello!\", \n",
" role=\"user\" \n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\", \n",
" content=\"hello!\", \n",
" ),\n",
" ]\n",
")\n",
"response"
]
@ -123,7 +121,7 @@
"id": "20a5ccf4-addd-4bdb-be80-161f7925dae0",
"metadata": {},
"source": [
"Note that MemGPT agents will generate an *internal_monologue* that explains its actions. You can use this monoloque to understand why agents are behaving as they are. \n",
"Note that MemGPT agents will generate a *reasoning_message* that explains its actions. You can use this monoloque to understand why agents are behaving as they are. \n",
"\n",
"Second, MemGPT agents also use tools to communicate, so messages are sent back by calling a `send_message` tool. This makes it easy to allow agent to communicate over different mediums (e.g. text), and also allows the agent to distinguish betweeh that is and isn't send to the end user. "
]
@ -175,7 +173,7 @@
"metadata": {},
"outputs": [],
"source": [
"memory = client.get_core_memory(agent_state.id)"
"memory = client.agents.core_memory.retrieve(agent_id=agent_state.id)"
]
},
{
@ -195,7 +193,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_archival_memory_summary(agent_state.id)"
"client.agents.context.retrieve(agent_id=agent_state.id)[\"num_archival_memory\"]"
]
},
{
@ -205,7 +203,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_recall_memory_summary(agent_state.id)"
"client.agents.context.retrieve(agent_id=agent_state.id)[\"num_recall_memory\"]"
]
},
{
@ -215,7 +213,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_messages(agent_state.id)"
"client.agents.messages.list(agent_id=agent_state.id)"
]
},
{
@ -243,11 +241,15 @@
"metadata": {},
"outputs": [],
"source": [
"response = client.send_message(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" message = \"My name is actually Bob\", \n",
" role = \"user\"\n",
") \n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\", \n",
" content=\"My name is actually Bob\", \n",
" ),\n",
" ]\n",
")\n",
"response"
]
},
@ -258,7 +260,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_core_memory(agent_state.id)"
"client.agents.core_memory.retrieve(agent_id=agent_state.id)"
]
},
{
@ -277,11 +279,15 @@
"metadata": {},
"outputs": [],
"source": [
"response = client.send_message(\n",
" agent_id=agent_state.id, \n",
" message = \"In the future, never use emojis to communicate\", \n",
" role = \"user\"\n",
") \n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id,\n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\", \n",
" content=\"In the future, never use emojis to communicate\", \n",
" ),\n",
" ]\n",
")\n",
"response"
]
},
@ -292,7 +298,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_core_memory(agent_state.id).get_block('persona')"
"client.agents.core_memory.retrieve_block(agent_id=agent_state.id, block_label='persona')"
]
},
{
@ -311,7 +317,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_archival_memory(agent_state.id)"
"client.agents.archival_memory.list(agent_id=agent_state.id)"
]
},
{
@ -321,7 +327,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_archival_memory_summary(agent_state.id)"
"client.agents.context.retrieve(agent_id=agent_state.id)[\"num_archival_memory\"]"
]
},
{
@ -339,11 +345,15 @@
"metadata": {},
"outputs": [],
"source": [
"response = client.send_message(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" message = \"Save the information that 'bob loves cats' to archival\", \n",
" role = \"user\"\n",
") \n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\", \n",
" content=\"Save the information that 'bob loves cats' to archival\", \n",
" ),\n",
" ]\n",
")\n",
"response"
]
},
@ -354,7 +364,7 @@
"metadata": {},
"outputs": [],
"source": [
"client.get_archival_memory(agent_state.id)[0].text"
"client.agents.archival_memory.list(agent_id=agent_state.id)[0].text"
]
},
{
@ -372,9 +382,9 @@
"metadata": {},
"outputs": [],
"source": [
"client.insert_archival_memory(\n",
" agent_state.id, \n",
" \"Bob's loves boston terriers\"\n",
"client.agents.archival_memory.create(\n",
" agent_id=agent_state.id, \n",
" text=\"Bob's loves boston terriers\"\n",
")"
]
},
@ -393,21 +403,17 @@
"metadata": {},
"outputs": [],
"source": [
"response = client.send_message(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" role=\"user\", \n",
" message=\"What animals do I like? Search archival.\"\n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\", \n",
" content=\"What animals do I like? Search archival.\", \n",
" ),\n",
" ]\n",
")\n",
"response"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "adc394c8-1d88-42bf-a6a5-b01f20f78d81",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {

4
examples/docs/agent_advanced.py
View File

@ -41,7 +41,7 @@ agent_state = client.agents.create(
print(f"Created agent with name {agent_state.name} and unique ID {agent_state.id}")
# message an agent as a user
response = client.agents.messages.send(
response = client.agents.messages.create(
agent_id=agent_state.id,
messages=[
MessageCreate(
@ -54,7 +54,7 @@ print("Usage", response.usage)
print("Agent messages", response.messages)
# message a system message (non-user)
response = client.agents.messages.send(
response = client.agents.messages.create(
agent_id=agent_state.id,
messages=[
MessageCreate(

4
examples/docs/agent_basic.py
View File

@ -24,7 +24,7 @@ agent_state = client.agents.create(
print(f"Created agent with name {agent_state.name} and unique ID {agent_state.id}")
# Message an agent
response = client.agents.messages.send(
response = client.agents.messages.create(
agent_id=agent_state.id,
messages=[
MessageCreate(
@ -40,7 +40,7 @@ print("Agent messages", response.messages)
agents = client.agents.list()
# get the agent by ID
agent_state = client.agents.get(agent_id=agent_state.id)
agent_state = client.agents.retrieve(agent_id=agent_state.id)
# get the agent by name
agent_state = client.agents.list(name=agent_state.name)[0]

115
examples/docs/example.py Normal file
View File

@ -0,0 +1,115 @@
from letta_client import CreateBlock, Letta, MessageCreate
"""
Make sure you run the Letta server before running this example.
```
letta server
```
Execute this script using `poetry run python3 example.py`
"""
client = Letta(
base_url="http://localhost:8283",
)
agent = client.agents.create(
memory_blocks=[
CreateBlock(
value="Name: Caren",
label="human",
),
],
model="openai/gpt-4o-mini",
embedding="openai/text-embedding-ada-002",
)
print(f"Created agent with name {agent.name}")
message_text = "What's my name?"
response = client.agents.messages.create(
agent_id=agent.id,
messages=[
MessageCreate(
role="user",
content=message_text,
),
],
)
print(f"Sent message to agent {agent.name}: {message_text}")
print(f"Agent thoughts: {response.messages[0].reasoning}")
print(f"Agent response: {response.messages[1].content}")
def secret_message():
"""Return a secret message."""
return "Hello world!"
tool = client.tools.upsert_from_function(
func=secret_message,
)
client.agents.tools.attach(agent_id=agent.id, tool_id=tool.id)
print(f"Created tool {tool.name} and attached to agent {agent.name}")
message_text = "Run secret message tool and tell me what it returns"
response = client.agents.messages.create(
agent_id=agent.id,
messages=[
MessageCreate(
role="user",
content=message_text,
),
],
)
print(f"Sent message to agent {agent.name}: {message_text}")
print(f"Agent thoughts: {response.messages[0].reasoning}")
print(f"Tool call information: {response.messages[1].tool_call}")
print(f"Tool response information: {response.messages[2].status}")
print(f"Agent thoughts: {response.messages[3].reasoning}")
print(f"Agent response: {response.messages[4].content}")
agent_copy = client.agents.create(
model="openai/gpt-4o-mini",
embedding="openai/text-embedding-ada-002",
)
block = client.agents.core_memory.retrieve_block(agent.id, "human")
agent_copy = client.agents.core_memory.attach_block(agent_copy.id, block.id)
print(f"Created agent copy with shared memory named {agent_copy.name}")
message_text = "My name isn't Caren, it's Sarah. Please update your core memory with core_memory_replace"
response = client.agents.messages.create(
agent_id=agent_copy.id,
messages=[
MessageCreate(
role="user",
content=message_text,
),
],
)
print(f"Sent message to agent {agent_copy.name}: {message_text}")
block = client.agents.core_memory.retrieve_block(agent_copy.id, "human")
print(f"New core memory for agent {agent_copy.name}: {block.value}")
message_text = "What's my name?"
response = client.agents.messages.create(
agent_id=agent_copy.id,
messages=[
MessageCreate(
role="user",
content=message_text,
),
],
)
print(f"Sent message to agent {agent_copy.name}: {message_text}")
print(f"Agent thoughts: {response.messages[0].reasoning}")
print(f"Agent response: {response.messages[1].content}")
client.agents.delete(agent_id=agent.id)
client.agents.delete(agent_id=agent_copy.id)
print(f"Deleted agents {agent.name} and {agent_copy.name}")

31
examples/docs/node/example.ts
View File

@ -6,7 +6,13 @@ import {
ToolReturnMessage,
} from '@letta-ai/letta-client/api/types';
// Start letta server and run `npm run example`
/**
* Make sure you run the Letta server before running this example.
* ```
* letta server
* ```
* Execute this script using `npm run example`
*/
const client = new LettaClient({
baseUrl: 'http://localhost:8283',
});
@ -56,9 +62,7 @@ const tool = await client.tools.upsert({
await client.agents.tools.attach(agent.id, tool.id!);
console.log(
`Created tool with name ${tool.name} and attached to agent ${agent.name}`,
);
console.log(`Created tool ${tool.name} and attached to agent ${agent.name}`);
messageText = 'Run secret message tool and tell me what it returns';
response = await client.agents.messages.create(agent.id, {
@ -70,21 +74,21 @@ response = await client.agents.messages.create(agent.id, {
],
});
console.log('Sent message to agent:', messageText);
console.log(`Sent message to agent ${agent.name}: ${messageText}`);
console.log(
'Agent thoughts',
'Agent thoughts:',
(response.messages[0] as ReasoningMessage).reasoning,
);
console.log(
'Tool call information',
'Tool call information:',
(response.messages[1] as ToolCallMessage).toolCall,
);
console.log(
'Tool response information',
'Tool response information:',
(response.messages[2] as ToolReturnMessage).status,
);
console.log(
'Agent thoughts',
'Agent thoughts:',
(response.messages[3] as ReasoningMessage).reasoning,
);
console.log(
@ -103,8 +107,6 @@ console.log('Created agent copy with shared memory named', agentCopy.name);
messageText =
"My name isn't Caren, it's Sarah. Please update your core memory with core_memory_replace";
console.log(`Sent message to agent ${agentCopy.name}: ${messageText}`);
response = await client.agents.messages.create(agentCopy.id, {
messages: [
{
@ -114,6 +116,8 @@ response = await client.agents.messages.create(agentCopy.id, {
],
});
console.log(`Sent message to agent ${agentCopy.name}: ${messageText}`);
block = await client.agents.coreMemory.retrieveBlock(agentCopy.id, 'human');
console.log(`New core memory for agent ${agentCopy.name}: ${block.value}`);
@ -136,3 +140,8 @@ console.log(
'Agent response:',
(response.messages[1] as AssistantMessage).content,
);
await client.agents.delete(agent.id);
await client.agents.delete(agentCopy.id);
console.log(`Deleted agents ${agent.name} and ${agentCopy.name}`);

2
examples/docs/rest_client.py
View File

@ -38,7 +38,7 @@ def main():
# Send a message to the agent
print(f"Created agent: {agent_state.name} with ID {str(agent_state.id)}")
response = client.agents.messages.send(
response = client.agents.messages.create(
agent_id=agent_state.id,
messages=[
MessageCreate(

14
examples/docs/tools.py
View File

@ -33,7 +33,7 @@ def roll_d20() -> str:
# create a tool from the function
tool = client.tools.upsert_from_function(func=roll_d20, name="roll_d20")
tool = client.tools.upsert_from_function(func=roll_d20)
print(f"Created tool with name {tool.name}")
# create a new agent
@ -59,7 +59,7 @@ agent_state = client.agents.create(
print(f"Created agent with name {agent_state.name} with tools {[t.name for t in agent_state.tools]}")
# Message an agent
response = client.agents.messages.send(
response = client.agents.messages.create(
agent_id=agent_state.id,
messages=[
MessageCreate(
@ -72,15 +72,15 @@ print("Usage", response.usage)
print("Agent messages", response.messages)
# remove a tool from the agent
client.agents.tools.remove(agent_id=agent_state.id, tool_id=tool.id)
client.agents.tools.detach(agent_id=agent_state.id, tool_id=tool.id)
# add a tool to the agent
client.agents.tools.add(agent_id=agent_state.id, tool_id=tool.id)
client.agents.tools.attach(agent_id=agent_state.id, tool_id=tool.id)
client.agents.delete(agent_id=agent_state.id)
# create an agent with only a subset of default tools
send_message_tool = client.tools.get_by_name(tool_name="send_message")
send_message_tool = [t for t in client.tools.list() if t.name == "send_message"][0]
agent_state = client.agents.create(
memory_blocks=[
CreateBlock(
@ -91,11 +91,11 @@ agent_state = client.agents.create(
model="openai/gpt-4o-mini",
embedding="openai/text-embedding-ada-002",
include_base_tools=False,
tool_ids=[tool.id, send_message_tool],
tool_ids=[tool.id, send_message_tool.id],
)
# message the agent to search archival memory (will be unable to do so)
client.agents.messages.send(
client.agents.messages.create(
agent_id=agent_state.id,
messages=[
MessageCreate(

21
examples/notebooks/Agentic RAG with Letta.ipynb
View File

@ -141,7 +141,6 @@
"client.sources.attach(\n",
" source_id=source.id,\n",
" agent_id=agent_state.id\n",
" \n",
")"
]
},
@ -241,7 +240,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id,\n",
" messages=[\n",
" MessageCreate(\n",
@ -408,7 +407,7 @@
"metadata": {},
"outputs": [],
"source": [
"agent_state = client.create_agent(\n",
"agent_state = client.agents.create(\n",
" name=\"birthday_agent\", \n",
" tool_ids=[birthday_tool.id],\n",
" memory_blocks=[\n",
@ -523,7 +522,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id,\n",
" messages=[\n",
" MessageCreate(\n",
@ -680,7 +679,7 @@
"\n",
"\"\"\"\n",
"\n",
"agent_state = client.create_agent(\n",
"agent_state = client.agents.create(\n",
" name=\"search_agent\", \n",
" memory_blocks=[\n",
" CreateBlock(\n",
@ -809,7 +808,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" messages=[\n",
" MessageCreate(\n",
@ -839,7 +838,7 @@
"from letta.schemas.llm_config import LLMConfig\n",
"\n",
"\n",
"agent_state = client.create_agent(\n",
"agent_state = client.agents.create(\n",
" name=\"search_agent\", \n",
" memory_blocks=[\n",
" CreateBlock(\n",
@ -958,14 +957,6 @@
")\n",
"response"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "91192bb7-4a74-4c94-a485-883d930b0489",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {

143
examples/notebooks/Customizing memory management.ipynb
View File

@ -248,7 +248,7 @@
"id": "fbdc9b6e-8bd5-4c42-970e-473da4adb2f2",
"metadata": {},
"source": [
"### Defining a memory module\n"
"### Defining task related tools\n"
]
},
{
@ -258,56 +258,46 @@
"metadata": {},
"outputs": [],
"source": [
"from letta import ChatMemory, Block \n",
"from typing import Optional, List\n",
"import json\n",
"\n",
"class TaskMemory(ChatMemory): \n",
"def task_queue_push(self: \"Agent\", task_description: str):\n",
" \"\"\"\n",
" Push to a task queue stored in core memory. \n",
"\n",
" def __init__(self, human: str, persona: str, tasks: List[str]): \n",
" super().__init__(human=human, persona=persona, limit=2000) \n",
" self.link_block( \n",
" Block(\n",
" limit=2000, \n",
" value=json.dumps(tasks), \n",
" label=\"tasks\"\n",
" )\n",
" )\n",
" Args:\n",
" task_description (str): A description of the next task you must accomplish. \n",
" \n",
" Returns:\n",
" Optional[str]: None is always returned as this function \n",
" does not produce a response.\n",
" \"\"\"\n",
" import json\n",
" tasks = json.loads(self.memory.get_block(\"tasks\").value)\n",
" tasks.append(task_description)\n",
" self.memory.update_block_value(\"tasks\", json.dumps(tasks))\n",
" return None\n",
"\n",
" def task_queue_push(self: \"Agent\", task_description: str):\n",
" \"\"\"\n",
" Push to a task queue stored in core memory. \n",
"def task_queue_pop(self: \"Agent\"):\n",
" \"\"\"\n",
" Get the next task from the task queue \n",
"\n",
" Args:\n",
" task_description (str): A description of the next task you must accomplish. \n",
" \n",
" Returns:\n",
" Optional[str]: None is always returned as this function \n",
" does not produce a response.\n",
" \"\"\"\n",
" import json\n",
" tasks = json.loads(self.memory.get_block(\"tasks\").value)\n",
" tasks.append(task_description)\n",
" self.memory.update_block_value(\"tasks\", json.dumps(tasks))\n",
" Returns:\n",
" Optional[str]: The description of the task popped from the \n",
" queue, if there are still tasks in queue. Otherwise, returns\n",
" None (the task queue is empty)\n",
" \"\"\"\n",
" import json\n",
" tasks = json.loads(self.memory.get_block(\"tasks\").value)\n",
" if len(tasks) == 0: \n",
" return None\n",
" task = tasks[0]\n",
" print(\"CURRENT TASKS: \", tasks)\n",
" self.memory.update_block_value(\"tasks\", json.dumps(tasks[1:]))\n",
" return task\n",
"\n",
" def task_queue_pop(self: \"Agent\"):\n",
" \"\"\"\n",
" Get the next task from the task queue \n",
" \n",
" Returns:\n",
" Optional[str]: The description of the task popped from the \n",
" queue, if there are still tasks in queue. Otherwise, returns\n",
" None (the task queue is empty)\n",
" \"\"\"\n",
" import json\n",
" tasks = json.loads(self.memory.get_block(\"tasks\").value)\n",
" if len(tasks) == 0: \n",
" return None\n",
" task = tasks[0]\n",
" print(\"CURRENT TASKS: \", tasks)\n",
" self.memory.update_block_value(\"tasks\", json.dumps(tasks[1:]))\n",
" return task\n"
"push_task_tool = client.tools.upsert_from_function(func=task_queue_push)\n",
"pop_task_tool = client.tools.upsert_from_function(func=task_queue_pop)"
]
},
{
@ -328,17 +318,28 @@
"task_agent_name = \"task_agent\"\n",
"\n",
"# delete agent if exists \n",
"if client.get_agent_id(task_agent_name): \n",
" client.delete_agent(client.get_agent_id(task_agent_name))\n",
"agents = client.agents.list(name=task_agent_name)\n",
"if len(agents) > 0: \n",
" client.agents.delete(agent_id=agents[0].id)\n",
"\n",
"task_agent_state = client.create_agent(\n",
"task_agent_state = client.agents.create(\n",
" name=task_agent_name, \n",
" system = open(\"data/task_queue_system_prompt.txt\", \"r\").read(),\n",
" memory=TaskMemory(\n",
" human=\"My name is Sarah\", \n",
" persona=\"You are an agent that must clear its tasks.\", \n",
" tasks=[]\n",
" )\n",
" memory_blocks=[\n",
" CreateBlock(\n",
" label=\"human\",\n",
" value=\"My name is Sarah\",\n",
" ),\n",
" CreateBlock(\n",
" label=\"persona\",\n",
" value=\"You are an agent that must clear its tasks.\",\n",
" ),\n",
" CreateBlock(\n",
" label=\"tasks\",\n",
" value=\"\",\n",
" ),\n",
" ],\n",
" tool_ids=[push_task_tool.id, pop_task_tool.id],\n",
")"
]
},
@ -491,10 +492,14 @@
}
],
"source": [
"response = client.send_message(\n",
"response = client.agents.messages.create(\n",
" agent_id=task_agent_state.id, \n",
" role=\"user\", \n",
" message=\"Add 'start calling me Charles' and 'tell me a haiku about my name' as two separate tasks.\"\n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\",\n",
" content=\"Add 'start calling me Charles' and 'tell me a haiku about my name' as two separate tasks.\",\n",
" )\n",
" ],\n",
")\n",
"response"
]
@ -580,10 +585,14 @@
}
],
"source": [
"response = client.send_message(\n",
"response = client.agents.messages.create(\n",
" agent_id=task_agent_state.id, \n",
" role=\"user\", \n",
" message=\"complete your tasks\"\n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\",\n",
" content=\"complete your tasks\",\n",
" )\n",
" ],\n",
")\n",
"response"
]
@ -669,10 +678,14 @@
}
],
"source": [
"response = client.send_message(\n",
"response = client.agents.messages.create(\n",
" agent_id=task_agent_state.id, \n",
" role=\"user\", \n",
" message=\"keep going\"\\\n",
" messages=[\n",
" MessageCreate(\n",
" role=\"user\",\n",
" content=\"keep going\",\n",
" )\n",
" ],\n",
")\n",
"response"
]
@ -695,16 +708,8 @@
}
],
"source": [
"client.get_in_context_memory(task_agent_state.id).get_block(\"tasks\")"
"client.agents.core_memory.retrieve_block(agent_id=task_agent_state.id, block_label=\"tasks\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "bfb41f81-26e0-4bb7-8a49-b90a2e8b9ec6",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {

33
examples/notebooks/Introduction to Letta.ipynb
View File

@ -7,6 +7,7 @@
"source": [
"# Introduction to Letta\n",
"> Make sure you run the Letta server before running this example using `letta server`\n",
"\n",
"This lab will go over: \n",
"1. Creating an agent with Letta\n",
"2. Understand Letta agent state (messages, memories, tools)\n",
@ -68,7 +69,7 @@
"metadata": {},
"outputs": [],
"source": [
"agent_state = client.create_agent(\n",
"agent_state = client.agents.create(\n",
" name=agent_name, \n",
" memory_blocks=[\n",
" CreateBlock(\n",
@ -164,7 +165,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" messages=[\n",
" MessageCreate(\n",
@ -315,7 +316,7 @@
"metadata": {},
"outputs": [],
"source": [
"memory = client.agents.core_memory.get_blocks(agent_id=agent_state.id)"
"memory = client.agents.core_memory.retrieve(agent_id=agent_state.id)"
]
},
{
@ -357,7 +358,7 @@
}
],
"source": [
"client.agents.archival_memory.get_summary(agent_id=agent_state.id)"
"client.agents.context.retrieve(agent_id=agent_state.id)[\"num_archival_memory\"]"
]
},
{
@ -378,7 +379,7 @@
}
],
"source": [
"client.agents.recall_memory.get_summary(agent_id=agent_state.id)"
"client.agents.context.retrieve(agent_id=agent_state.id)[\"num_recall_memory\"]"
]
},
{
@ -524,7 +525,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" messages=[\n",
" MessageCreate(\n",
@ -554,7 +555,7 @@
}
],
"source": [
"client.agents.core_memory.get_blocks(agent_id=agent_state.id)"
"client.agents.core_memory.retrieve(agent_id=agent_state.id)"
]
},
{
@ -677,7 +678,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" messages=[\n",
" MessageCreate(\n",
@ -707,7 +708,7 @@
}
],
"source": [
"client.agents.core_memory.get_block(agent_id=agent_state.id, block_label='persona')"
"client.agents.core_memory.retrieve_block(agent_id=agent_state.id, block_label='persona')"
]
},
{
@ -758,7 +759,7 @@
}
],
"source": [
"client.agents.archival_memory.get_summary(agent_id=agent_state.id)"
"client.agents.context.retrieve(agent_id=agent_state.id)[\"num_archival_memory\"]"
]
},
{
@ -865,7 +866,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" messages=[\n",
" MessageCreate(\n",
@ -1034,7 +1035,7 @@
}
],
"source": [
"response = client.agents.messages.send(\n",
"response = client.agents.messages.create(\n",
" agent_id=agent_state.id, \n",
" messages=[\n",
" MessageCreate(\n",
@ -1045,14 +1046,6 @@
")\n",
"response"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "7c9b39df-d4ca-4d12-a6c4-cf3d0efa9738",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {

28
examples/notebooks/Multi-agent recruiting workflow.ipynb
View File

@ -123,7 +123,6 @@
" \"\"\"\n",
" import os\n",
" filepath = os.path.join(\"data\", \"resumes\", name.lower().replace(\" \", \"_\") + \".txt\")\n",
" #print(\"read\", filepath)\n",
" return open(filepath).read()\n",
"\n",
"def submit_evaluation(self, candidate_name: str, reach_out: bool, resume: str, justification: str): \n",
@ -154,8 +153,8 @@
"\n",
"# TODO: add an archival andidate tool (provide justification) \n",
"\n",
"read_resume_tool = client.tools.upsert_from_function(name=\"read_resume\", func=read_resume) \n",
"submit_evaluation_tool = client.tools.upsert_from_function(name=\"submit_evaluation\", func=submit_evaluation)"
"read_resume_tool = client.tools.upsert_from_function(func=read_resume) \n",
"submit_evaluation_tool = client.tools.upsert_from_function(func=submit_evaluation)"
]
},
{
@ -213,7 +212,7 @@
" print(\"Pretend to email:\", content)\n",
" return\n",
"\n",
"email_candidate_tool = client.tools.upsert_from_function(name=\"email_candidate\", func=email_candidate)"
"email_candidate_tool = client.tools.upsert_from_function(func=email_candidate)"
]
},
{
@ -668,7 +667,7 @@
}
],
"source": [
"client.get_block(org_block.id)"
"client.blocks.retrieve(block_id=org_block.id)"
]
},
{
@ -718,8 +717,8 @@
"\n",
"\n",
"# create tools \n",
"search_candidate_tool = client.tools.upsert_from_function(name=\"search_candidates_db\", func=search_candidates_db)\n",
"consider_candidate_tool = client.tools.upsert_from_function(name=\"consider_candidate\", func=consider_candidate)\n",
"search_candidate_tool = client.tools.upsert_from_function(func=search_candidates_db)\n",
"consider_candidate_tool = client.tools.upsert_from_function(func=consider_candidate)\n",
"\n",
"# create recruiter agent\n",
"recruiter_agent = client.agents.create(\n",
@ -855,18 +854,9 @@
"metadata": {},
"outputs": [],
"source": [
"client.agents.delete(eval_agent.id)\n",
"client.agents.delete(outreach_agent.id)"
]
},
{
"cell_type": "code",
"execution_count": 29,
"id": "672f941e-af17-4b5c-8a21-925a1d88c47f",
"metadata": {},
"outputs": [],
"source": [
"client.agents.delete(recruiter_agent.id)"
"client.agents.delete(agent_id=eval_agent.id)\n",
"client.agents.delete(agent_id=outreach_agent.id)\n",
"client.agents.delete(agent_id=recruiter_agent.id)"
]
}
],

13
examples/tutorials/memgpt_rag_agent.ipynb
View File

@ -49,7 +49,7 @@
"metadata": {},
"outputs": [],
"source": [
"letta_paper = client.create_source(\n",
"letta_paper = client.sources.create(\n",
" name=\"letta_paper\", \n",
")"
]
@ -69,7 +69,7 @@
"metadata": {},
"outputs": [],
"source": [
"job = client.load_file_to_source(filename=filename, source_id=letta_paper.id)\n",
"job = client.sources.files.upload(filename=filename, source_id=letta_paper.id)\n",
"job"
]
},
@ -89,14 +89,19 @@
"metadata": {},
"outputs": [],
"source": [
"client.attach_source_to_agent(source_id=letta_paper.id, agent_id=basic_agent.id)\n",
"client.agents.sources.attach(source_id=letta_paper.id, agent_id=basic_agent.id)\n",
"# TODO: add system message saying that file has been attached \n",
"\n",
"from pprint import pprint\n",
"\n",
"# TODO: do soemthing accenture related \n",
"# TODO: brag about query rewriting -- hyde paper \n",
"response = client.user_message(agent_id=basic_agent.id, message=\"what is core memory? search your archival memory.\") \n",
"response = client.agents.messages.create(agent_id=basic_agent.id, messages=[\n",
" MessageCreate(\n",
" role=\"user\",\n",
" content=\"what is core memory? search your archival memory.\",\n",
" )\n",
"])\n",
"pprint(response.messages)"
]
}

1
letta/agent.py
View File

@ -235,6 +235,7 @@ class Agent(BaseAgent):
# TODO: This is only temporary, can remove after we publish a pip package with this object
agent_state_copy = self.agent_state.__deepcopy__()
agent_state_copy.tools = []
agent_state_copy.tool_rules = []
sandbox_run_result = ToolExecutionSandbox(function_name, function_args, self.user).run(agent_state=agent_state_copy)
function_response, updated_agent_state = sandbox_run_result.func_return, sandbox_run_result.agent_state

6
letta/orm/custom_columns.py
View File

@ -84,11 +84,11 @@ class ToolRulesColumn(TypeDecorator):
def deserialize_tool_rule(data: dict) -> Union[ChildToolRule, InitToolRule, TerminalToolRule, ConditionalToolRule]:
"""Deserialize a dictionary to the appropriate ToolRule subclass based on the 'type'."""
rule_type = ToolRuleType(data.get("type")) # Remove 'type' field if it exists since it is a class var
if rule_type == ToolRuleType.run_first:
if rule_type == ToolRuleType.run_first or rule_type == "InitToolRule":
return InitToolRule(**data)
elif rule_type == ToolRuleType.exit_loop:
elif rule_type == ToolRuleType.exit_loop or rule_type == "TerminalToolRule":
return TerminalToolRule(**data)
elif rule_type == ToolRuleType.constrain_child_tools:
elif rule_type == ToolRuleType.constrain_child_tools or rule_type == "ToolRule":
rule = ChildToolRule(**data)
return rule
elif rule_type == ToolRuleType.conditional:

12
letta/schemas/enums.py
View File

@ -46,9 +46,13 @@ class ToolRuleType(str, Enum):
# note: some of these should be renamed when we do the data migration
run_first = "InitToolRule"
exit_loop = "TerminalToolRule" # reasoning loop should exit
continue_loop = "continue_loop" # reasoning loop should continue
run_first = "run_first"
exit_loop = "exit_loop" # reasoning loop should exit
continue_loop = "continue_loop"
conditional = "conditional"
constrain_child_tools = "ToolRule"
constrain_child_tools = "constrain_child_tools"
require_parent_tools = "require_parent_tools"
# Deprecated
InitToolRule = "InitToolRule"
TerminalToolRule = "TerminalToolRule"
ToolRule = "ToolRule"

2
letta/schemas/tool_rule.py
View File

@ -9,7 +9,7 @@ from letta.schemas.letta_base import LettaBase
class BaseToolRule(LettaBase):
__id_prefix__ = "tool_rule"
tool_name: str = Field(..., description="The name of the tool. Must exist in the database for the user's organization.")
type: ToolRuleType
type: ToolRuleType = Field(..., description="The type of the message.")
class ChildToolRule(BaseToolRule):

1
tests/integration_test_tool_execution_sandbox.py
View File

@ -232,6 +232,7 @@ def agent_state():
embedding_config=EmbeddingConfig.default_config(provider="openai"),
llm_config=LLMConfig.default_config(model_name="gpt-4"),
)
agent_state.tool_rules = []
yield agent_state