LangSmith API. HTTPError('403 Client Error: Forbidden for url: https://api.smith.langchain.com/runs/multipart', '{"error":"Forbidden"}\n')

[stephanyvargas]

Issue you'd like to raise.

Background (No idea if it is relevant): I have a personal account and an organizational account with my company under the same email.

1. Problem: I am trying to get the onboarding to work but keep hitting 403s everywhere.

2. System run: using uv paired with python

3. My system dependencies (toml):

dependencies = [
    "elevenlabs>=2.27.0",
    "ipython>=9.8.0",
    "langchain>=1.1.3",
    "langchain-openai>=1.1.3",
    "langgraph>=1.0.5",
    "matplotlib>=3.10.8",
    "numpy>=2.3.5",
    "python-dotenv>=1.2.1",
    "scikit-learn>=1.8.0",
    "websockets>=15.0.1",
    "aiofiles>=25.1.0",
    "aiohttp[speedups]>=3.13.3",
    "dateparser>=1.3.0",
    "fastapi>=0.129.0",
    "gunicorn>=25.1.0",
    "httpx[http2]>=0.28.1",
    "pydantic>=2.12.5",
    "pydantic-core>=2.41.5",
    "requests>=2.32.5",
    "structlog>=25.5.0",
    "uvicorn[standard]>=0.41.0",
    "langsmith>=0.7.7",
]

$ uv pip show langsmith
Name: langsmith
Version: 0.7.7
Location: /home/.venv/lib/python3.12/site-packages
Requires: httpx, orjson, packaging, pydantic, requests, requests-toolbelt, uuid-utils, xxhash, zstandard
Required-by: langchain-core

4. The.env file looks like this:

# LLM API key
OPENAI_API_KEY=sk-proj-redacted
# LangSmith Configuration
LANGSMITH_TRACING=true
LANGSMITH_ENDPOINT=https://api.smith.langchain.com -> I double checked, this is correct!
LANGSMITH_API_KEY=lsv2_pt_redacted
LANGSMITH_PROJECT=development_conversation_agent
LANGSMITH_WORKSPACE_ID=4c112d07-redacted -> I triple check for the project and workspace UUID, actually tried both, none worked

5. The offending code

import os
from dotenv import load_dotenv
load_dotenv()

from langchain.agents import create_agent

print(os.getenv('LANGSMITH_API_KEY')[:10]) -> this is printing!

def get_weather(city: str) -> str:
    """Get weather for a given city."""
    return f"It's always sunny in {city}!"


agent = create_agent(
    model="openai:gpt-5-mini",
    tools=[get_weather],
    system_prompt="You are a helpful assistant",
)

# Run the agent
agent.invoke(
    {"messages": [{"role": "user", "content": "What is the weather in San Francisco?"}]}
    )

6. run as uv run python test_lansmith_connection.py, exact output from terminal

lsv2_sk_ca
Failed to multipart ingest runs: langsmith.utils.LangSmithError: Failed to POST https://api.smith.langchain.com/runs/multipart in LangSmith API. HTTPError('403 Client Error: Forbidden for url: https://api.smith.langchain.com/runs/multipart', '{"error":"Forbidden"}\n')
Failed to send compressed multipart ingest: langsmith.utils.LangSmithError: Failed to POST https://api.smith.langchain.com/runs/multipart in LangSmith API. HTTPError('403 Client Error: Forbidden for url: https://api.smith.langchain.com/runs/multipart', '{"error":"Forbidden"}\n')
Failed to send compressed multipart ingest: langsmith.utils.LangSmithError: Failed to POST https://api.smith.langchain.com/runs/multipart in LangSmith API. HTTPError('403 Client Error: Forbidden for url: https://api.smith.langchain.com/runs/multipart', '{"error":"Forbidden"}\n')

7. Used a key generated on the fly during onborading session (didn't work), tried service key and manually creating a key from the workspace (didn't work)

8. Last step: spend hours fighting with Chatgpt, Claude, polly and random fixes from the internet with no success =C

Suggestion:

No response

[bvs-langchain]

Hi Stephany Vargas (@stephanyvargas), could you please try this without specifying LANGSMITH_WORKSPACE_ID in the environment? PATs have a default workspace which will be used in the absence of this configuration.

If that doesn't work, please reach out to support@langchain.dev with information including the workspace ID you're attempting to submit traces to, as well as the "Short key" which you can find in your list of API keys, formatted like lsv2_pt_8...fe4c

[stephanyvargas]

What worked for me (using uv in a package project),

1. .env file with the following entries (NO LANGSMITH_WORKSPACE_ID for the latest releases, double check packages)

# LLM configuration 
OPENAI_API_KEY=sk-redacted
# LangSmith Configuration
LANGSMITH_TRACING=true
LANGSMITH_ENDPOINT=https://api.smith.langchain.com
LANGSMITH_API_KEY=lsv2_pt_redacted
LANGSMITH_PROJECT=development_redacted

Then source .env

2. Installation (Note the --pre part)
   pip install --pre -U langchain langchain-openai

3. Find the versions of the following packages
   langchain, langchain-core and langchain-openai

4. In the pyproject.toml: Add those packages to the dependencies part BEFORE langsmith installation and allow pre-releases

dependencies = [
    "langchain>=1.2.10",
    "langchain-core>=1.2.16",
    "langchain-openai>=1.1.10",
    "langgraph>=1.0.10rc1",
    "langsmith>=0.7.7",
...

[tool.uv]
prerelease = "allow"

5. Update uv -> uv sync --prerelease=allow

6. Test if it worked -> uv run python test_langsmith_debug.py
   python code

# python test_langsmith_debug.py

import os
from dotenv import load_dotenv
load_dotenv()

print("=== LANGSMITH envs visible to Python ===")
for k in sorted(os.environ):
    if k.startswith("LANGSMITH"):
        print(k, "=", repr(os.environ[k]))

print("\n=== Testing LangSmith Client() ===")
from langsmith import Client

client = Client()
try:
    projects = list(client.list_projects(limit=3))
    print("OK, got projects:")
    for p in projects:
        print(" -", getattr(p, "name", p))
except Exception as e:
    print("ERROR:", repr(e))

Hope this helps someone =D