Unable to retrieve organization_id | create_chat request error

[kqvanity]

when I try to automatically login using selenium, i get the following error (I have two profile at firefox btw, and I sign claude into both of them)

  python3 unofficial-claude-dir/unofficial-claude.py

Retrieving Claude session cookie from /home/keinvanity/.mozilla/firefox/gmpc627g.default-release

Unable to retrieve organization_id from profile: /home/keinvanity/.mozilla/firefox/gmpc627g.default-release
Check if this profile is logged into Claude!
Traceback (most recent call last):
  File "/home/keinvanity/Downloads/UnofficialClaude/unofficial-claude-dir/unofficial-claude.py", line 32, in <module>
    client = ClaudeAPIClient(session, model_name='claude-3-5-sonnet-20240620', timeout=240)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/keinvanity/Downloads/UnofficialClaude/claude/lib/python3.12/site-packages/claude_api/client.py", line 147, in __init__
    raise ValueError("Invalid SessionData argument!")
ValueError: Invalid SessionData argument!

when omitting selenium, and provide the organization id, and session key manually, I get this error instead

  python3 unofficial-claude-dir/unofficial-claude.py

Message limit hit, cannot create chat...

[st1vms]

@kqvanity Can you please post the result of visiting https://claude.ai/api/organizations through logged Firefox profile?
Maybe it is related to your default organization being unavailable, and this could be a duplicate of #23, just to check...

As I see it, get_session_data failed to parse the organization uuid from that endpoint...And as a result it returned None which throwed the exception on ClaudeAPIClient. This usually means the json returned from /api/organizations was invalid...

[kqvanity]

post the result of visiting

[
  {
    "id": 14182401,
    "uuid": "552779e3-b45a-4d48-9e05-6d152ca782ea",
    "name": "Github",
    "settings": {
      "claude_console_privacy": "default_private",
      "allowed_invite_domains": null
    },
    "capabilities": [
      "api"
    ],
    "rate_limit_tier": "auto_api_evaluation",
    "billing_type": null,
    "free_credits_status": null,
    "data_retention": "default",
    "api_disabled_reason": "out_of_credits",
    "api_disabled_until": null,
    "billable_usage_paused_until": null,
    "raven_type": null,
    "created_at": "2024-06-21T17:03:47.830914Z",
    "updated_at": "2024-06-21T17:03:47.830914Z",
    "active_flags": []
  },
  {
    "id": 23162662,
    "uuid": "3675554f-5b05-4f44-881f-b66e28869268",
    "name": "solicitmind@gmail.com's Organization",
    "settings": {
      "claude_console_privacy": "default_private",
      "allowed_invite_domains": null
    },
    "capabilities": [
      "chat"
    ],
    "rate_limit_tier": "default_claude_ai",
    "billing_type": null,
    "free_credits_status": null,
    "data_retention": "default",
    "api_disabled_reason": null,
    "api_disabled_until": null,
    "billable_usage_paused_until": null,
    "raven_type": null,
    "created_at": "2024-09-29T09:24:20.496864Z",
    "updated_at": "2024-09-29T09:24:20.496864Z",
    "active_flags": []
  }
]

could be duplicate of

I did check it out before, but knowing that the 'solution' got merged, I thought that ought to be a different error

[st1vms]

I did check it out before, but knowing that the 'solution' got merged, I thought that ought to be a different error

Did you retrieved this with the profile that's causing errors?
It seems you do also have the last organization set with capabilities chat.

I guess using get_session_data(organization_index=1) (or with 0) wouldn't change much of a thing tho...as the intended behaviour should look for the last entry in that json...but still it can't find the uuid in there...
I would need to inspect the json retrieved here

unofficial-claude-api/claude_api/session.py

Line 85 in 6447d83

j = json_loads(pre.text)

If you could install this library from source and print the json on that line when the error occurs it would be very useful.

[st1vms]

You can also inspect if selenium starts with a logged profile by turning off headless mode in this line:

unofficial-claude-api/claude_api/session.py

Line 63 in 6447d83

opts = get_firefox_options(firefox_profile=profile, headless=True)

My first assumption would be that the profile is not logged...

[kqvanity]

@st1vms I keep changing those lines at the the virtual environment's dir at nvim venv/lib/python3.12/site-packages/claude_api/session.py , but no changes are visible!

[st1vms]

@st1vms I keep changing those lines at the the virtual environment's dir at nvim venv/lib/python3.12/site-packages/claude_api/session.py , but no changes are visible!

I suggest you try outside virtual environment, by installing from source using

pip install -e .

with a terminal inside the source dir.

[kqvanity]

sorry. I had to let the script auto fetch the credentials with selenium for it to pick the changes I've made (instead of hardcoding them). Here's the output.

  python3 unofficial-claude.py

Retrieving Claude session cookie from /home/keinvanity/.mozilla/firefox/gmpc627g.default-release
pre {"type":"error","error":{"type":"permission_error","message":"Invalid authorization"}}
j {'type': 'error', 'error': {'type': 'permission_error', 'message': 'Invalid authorization'}}

Unable to retrieve organization_id from profile: /home/keinvanity/.mozilla/firefox/gmpc627g.default-release
Check if this profile is logged into Claude!
Traceback (most recent call last):
  File "/home/keinvanity/Downloads/claude-misc/unofficial-claude-dir/unofficial-claude.py", line 33, in <module>
    client = ClaudeAPIClient(session, model_name='claude-3-5-sonnet-20240620', timeout=240)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/keinvanity/Downloads/claude-misc/unofficial-claude-dir/venv/lib/python3.12/site-packages/claude_api/client.py", line 147, in __init__
    raise ValueError("Invalid SessionData argument!")
ValueError: Invalid SessionData argument!

It's also worth mentioning that selenium opens up firefox with old extensions of mine still being present! I can't persist firefox to inspect it.

[kqvanity]

Selenium basically loads another profile that doesn't have claude logged in. I don't know how to set which Firefox profile to load

[st1vms]

Does it work when not giving any profile to get_session_data ?

Apart from double checking the profile path...
I don't know...do you have the latest geckodriver executable and selenium library installed? Maybe you need to upgrade them...

[kqvanity]

Doe it work when not giving any profile

I already tried to pass the values right away. I visited https://claude.ai/api/organizations. Copied the second UUID, request cookie starting with CH-prefers-color-sche and then hardcoded those values in the example.py file you've provided at the readme, only to be prompted with

Message limit hit, cannot create chat..

[st1vms]

Doe it work when not giving any profile

I already tried to pass the values right away. I visited https://claude.ai/api/organizations. Copied the second UUID, request cookie starting with CH-prefers-color-sche and then hardcoded those values in the example.py file you've provided at the readme, only to be prompted with

Message limit hit, cannot create chat..

I'm sorry but I can't recreate the issue on my end.

Can you please take a look at the response in create_chat when giving all parameters manually (cookie, user_agent, organization_id)?
You should insert a print(response.content) in this line:

unofficial-claude-api/claude_api/client.py

Line 338 in 6447d83

if response and response.status_code == 201:

[kqvanity]

That's the change I've made

        print(payload)
        print(headers)
        response = http_post(
            url,
            headers=headers,
            data=payload,
            proxies=self.__get_proxy(),
            timeout=self.timeout,
            impersonate="chrome110",
        )
        print(response.text)
        if response and response.status_code == 201:
            j = response.json()
            if j and "uuid" in j:
                return j["uuid"]
        return None

and that's the output

{"name":"","uuid":"ae0cc...............-719249bfd0a4"}
{'Host': 'claude.ai', 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/94.0.4606.81 Safari/537.36', 'Accept': '*/*', 'Accept-Language': 'en-US,en;q=0.5', 'Accept-Encoding': 'gzip, deflate, br', 'Content-Type': 'application/json', 'Content-Length': '57', 'Referer': 'https://claude.ai/chats', 'Origin': 'https://claude.ai', 'DNT': '1', 'Sec-Fetch-Dest': 'empty', 'Sec-Fetch-Mode': 'cors', 'Sec-Fetch-Site': 'same-origin', 'Connection': 'keep-alive', 'Cookie': 'CH-prefers-color-scheme=light; __ssid=a5addfdf8.......f8e1438aa80; lastActiveOrg=3675554f-5b.......-b66e28869268; activitySessionId=02850caa-ec.............8a-297805bbbccd; __stripe_mid=cad07f96-fc54-4c73-89b0-e2f239331fac1ff7ad; sessionKey=sk-ant-sid01...................27A', 'TE': 'trailers'}
{"type":"error","error":{"type":"invalid_request_error","message":"Input should be a valid dictionary o


Message limit hit, cannot create chat...

[st1vms]

The error indicates a problem with the payload, but I can not recreate the bug on my end.

Maybe try with another account and see if it works.

I doubt they changed that endpoint...