Tools using "agent_state" result in ModuleNotFoundError

[SootyOwl]

Describe the bug
When trying to use the agent_state argument in a tool, the following error is always returned:

Error executing function create_research_plan: ModuleNotFoundError: No module named 'letta'

Details

---------------------------------------------------------------------------
ModuleNotFoundError
Traceback (most recent call last)
Cell In[1], line 12
     10 import struct
     11 import hashlib
---> 12 import letta
     13 from letta import *
     14 agent_state = pickle.loads(...)  # truncated
ModuleNotFoundError: No module named 'letta'

Steps to replicate
Create a tool which accepts the agent_state parameter as described in the documentation, attach it to an agent and ask the agent to run it.

Please describe your setup

• How are you running Letta?
  • Docker compose, letta v0.13
• Describe your setup
  • Linux

Screenshots

[cpfiffer]

This tool may be old, and is not overwritten when a new agent is created. Can you open the create_research_plan tool in the tool manager? you should be able to replace it with something like this:

def create_research_plan(agent_state: "AgentState", research_plan: List[str], topic: str):
    import json

    research_plan_str = f"""The plan of action is to research `{topic}` with the following steps: \n"""
    for i, step in enumerate(research_plan):
        research_plan_str += f"- [ ] Step {i+1} - {step}\n"

    agent_state.memory.update_block_value(label="research_plan", value=research_plan_str)

    return "Research plan successfully created, time to execute the plan!"

You'll need to update the schema as well, which should look like:

{
  "name": "create_research_plan",
  "description": "Think carefully about your research plan and come up with 3-5 research steps (provided in the research_plan list). Each step should describe that topics
 or questions to research during that research step. \n\nExamples: \ncreate_search_plan(['Top developing countries by GDP', 'Determine smartphone penetration share by
country', 'Identify langugage learning popularity across countries']) \ncreate_search_plan(['Search for AI SDR companies', 'Identify company size and funding', 'Search
for comparisons or competitive analysis for each company's product'])",
  "parameters": {
    "type": "object",
    "properties": {
      "research_plan": {
        "type": "array",
        "items": {
          "type": "string"
        },
        "description": "The sequential research plan to help guide the search process"
      },
      "topic": {
        "type": "string",
        "description": "The research topic"
      }
    },
    "required": [
      "research_plan",
      "topic"
    ]
  }
}

[SootyOwl]

That's already how the tool reads, but I replaced it verbatim with what you provided anyway and still got the same error. My docker logs show the following:

letta          | e2b.api - INFO - Request POST https://api.e2b.app/sandboxes
letta          | INFO:     192.168.1.202:57726 - "GET /v1/agents/agent-39232921-7b69-493f-9347-a783a62cde41 HTTP/1.1" 200 OK
letta          | httpx - INFO - HTTP Request: POST https://api.e2b.app/sandboxes "HTTP/1.1 201 Created"
letta          | e2b.api - INFO - Response 201
letta          | Letta.letta.services.tool_sandbox.e2b_sandbox - INFO - E2B Sandbox configurations: created_by_id='user-00000000-0000-4000-8000-000000000000' last_updated_by_id='user-00000000-0000-4000-8000-000000000000' created_at=datetime.datetime(2025, 7, 29, 23, 7, 11, 734354, tzinfo=datetime.timezone.utc) updated_at=datetime.datetime(2025, 10, 12, 14, 46, 24, 541509, tzinfo=datetime.timezone.utc) id='sandbox-93048d62-6917-4281-8b6b-9fb57f25ca31' type=<SandboxType.LOCAL: 'local'> organization_id='org-00000000-0000-4000-8000-000000000000' config={'sandbox_dir': '/root/.letta/tool_execution_dir', 'use_venv': False, 'venv_name': 'venv', 'pip_requirements': []}
letta          | Letta.letta.services.tool_sandbox.e2b_sandbox - INFO - E2B Sandbox ID: ihiuae07peupeyk5u2gck
letta          | Letta.letta.services.tool_sandbox.e2b_sandbox - INFO - E2B execution started for ID ihiuae07peupeyk5u2gck: create_research_plan
letta          | e2b.sandbox_async.main - INFO - Request: POST https://49999-ihiuae07peupeyk5u2gck.e2b.app/execute
letta          | e2b.sandbox_async.main - INFO - Response: 200 https://49999-ihiuae07peupeyk5u2gck.e2b.app/execute
letta          | httpx - INFO - HTTP Request: POST https://49999-ihiuae07peupeyk5u2gck.e2b.app/execute "HTTP/1.1 200 OK"
letta          | INFO:     172.22.0.1:41268 - "GET /v1/agents/agent-39232921-7b69-493f-9347-a783a62cde41/context HTTP/1.1" 200 OK
letta          | INFO:     192.168.1.203:50844 - "GET /v1/runs/?agent_id=agent-39232921-7b69-493f-9347-a783a62cde41&active=true HTTP/1.1" 200 OK
letta          | Letta.letta.services.tool_sandbox.e2b_sandbox - INFO - E2B execution failed for ID ihiuae07peupeyk5u2gck: create_research_plan
letta          | Letta.letta.services.tool_sandbox.e2b_sandbox - INFO - E2B sandbox ihiuae07peupeyk5u2gck killed
letta          | e2b.api - INFO - Request DELETE https://api.e2b.app/sandboxes/ihiuae07peupeyk5u2gck
letta          | httpx - INFO - HTTP Request: DELETE https://api.e2b.app/sandboxes/ihiuae07peupeyk5u2gck "HTTP/1.1 204 No Content"
letta          | e2b.api - INFO - Response 204
letta          | Letta.agent-39232921-7b69-493f-9347-a783a62cde41 - INFO - Running final update. Step Progression: StepProgression.FINISHED