search

gptscript-ai / gptscript

Build AI assistants that interact with your systems
https://gptscript.ai
Apache License 2.0
2.89k stars 256 forks source link

UI - Prompted for entering openai keys when launched with other model providers. #559

Closed sangee2004 closed 2 months ago

sangee2004 commented 2 months ago

Server - gptscript version v0.8.5-rc4+3033b05a

Steps to reproduce the problem:

  1. Make sure you do not have openai key set as env variable or credential (sys.openai) set.
  2. Have github.com/gptscript-ai/claude3-anthropic-provider/credential set.
  3. Launching UI with default provider set to Anthropic provider results in user being prompted for openai api key. Screenshot 2024-06-25 at 6 25 48 PM

When u enter an invalid value, it gets stuck and following errors are seen in console

 ⨯ unhandledRejection: error, status code: 401, message: Incorrect API key provided: asdf. You can find your API key at https://platform.openai.com/account/api-keys.
 ⨯ unhandledRejection: error, status code: 401, message: Incorrect API key provided: asdf. You can find your API key at https://platform.openai.com/account/api-keys.

Tried to launch UI by any of the 2 means results in the same behavior:

  1. by passing --default-model - gptscript --ui --disable-cache --default-model 'claude-3-5-sonnet-20240620 from [github.com/gptscript-ai/claude3-anthropic-provider](http://github.com/gptscript-ai/claude3-anthropic-provider)' --ui [github.com/gptscript-ai/cli-demo](http://github.com/gptscript-ai/cli-demo) or
  2. by setting ENV variable GPTSCRIPT_DEFAULT_MODEL=claude-3-5-sonnet-20240620 from github.com/gptscript-ai/claude3-anthropic-provider.

Logs:

gptscript --ui --disable-cache --ui github.com/gptscript-ai/cli-demo
18:23:16 WARNING: Changing the default model can have unknown behavior for existing tools. Use the model field per tool instead.
18:23:18 started  [main] [input=--file=github.com/gptscript-ai/cli-demo]
18:23:18 started  [context: github.com/gptscript-ai/context/os]
18:23:18 sent     [context: github.com/gptscript-ai/context/os]
18:23:18 ended    [context: github.com/gptscript-ai/context/os] [output=The local operating systems is Darwin, release 23.3.0]
18:23:20 started  [provider: https://raw.githubusercontent.com/gptscript-ai/claude3-anthropic-provider/5b581e7b84bdd99f5d56df593d99397c9f91a8e3/tool.gpt:Anthropic Claude3 Model Provider]
18:23:20 launched [Anthropic Claude3 Model Provider][https://raw.githubusercontent.com/gptscript-ai/claude3-anthropic-provider/5b581e7b84bdd99f5d56df593d99397c9f91a8e3/tool.gpt:Anthropic Claude3 Model Provider] port [10359] [/opt/homebrew/bin/gptscript sys.daemon /usr/bin/env python3 /Users/sangeethahariharan/Library/Caches/gptscript/repos/5b581e7b84bdd99f5d56df593d99397c9f91a8e3/tool.gpt/python3.12/main.py]
18:23:21 ended    [provider: https://raw.githubusercontent.com/gptscript-ai/claude3-anthropic-provider/5b581e7b84bdd99f5d56df593d99397c9f91a8e3/tool.gpt:Anthropic Claude3 Model Provider] [output=http://127.0.0.1:10359]
18:23:21 sent     [main]
         content  [1] content | Waiting for model response...
         content  [1] content | <tool call> service -> null
         content  [2] content | The local operating systems is Darwin, release 23.3.0
         content  [2] content | 
18:23:22 started  [service(4)] [input=null]
18:23:22 launched [service][https://raw.githubusercontent.com/gptscript-ai/ui/a5946810bc100357327cc11c1548d6bf7e408a09/tool.gpt:service] port [10993] [/opt/homebrew/bin/gptscript sys.daemon /usr/bin/env npm run --prefix /Users/sangeethahariharan/Library/Caches/gptscript/repos/a5946810bc100357327cc11c1548d6bf7e408a09/tool.gpt/node21 dev]

> next-app-template@0.0.1 dev
> node server.mjs

> Socket server is ready at http://localhost:10993
 ○ Compiling / ...
 ✓ Compiled / in 2.1s (4300 modules)
 GET / 200 in 2374ms
18:23:27 ended    [service(4)] [output=\u003c!DOCTYPE html\u003e\u003chtml lang=\"en\"\u003e\u003chead\u003e\u003cmeta charSet=\"utf-8\"/\u003e\u003cmeta name=\"viewport\" content=\"width=dev...]
18:23:27 continue [main]
 POST / 200 in 16ms
18:23:27 started  [context: github.com/gptscript-ai/context/os]
18:23:27 sent     [context: github.com/gptscript-ai/context/os]
18:23:27 ended    [context: github.com/gptscript-ai/context/os] [output=The local operating systems is Darwin, release 23.3.0]
18:23:27 sent     [main]
         content  [1] content | Waiting for model response...
         content  [1] content | <tool call> port -> null         content  [5] content | The local operating systems is Darwin, release 23.3.0
         content  [5] content | 

18:23:29 started  [port(6)] [input=null]
 ✓ Compiled /api/port in 218ms (2304 modules)
 POST /api/port 200 in 256ms
18:23:30 ended    [port(6)] [output=10993]
18:23:30 continue [main]
18:23:30 started  [context: github.com/gptscript-ai/context/os]
18:23:30 sent     [context: github.com/gptscript-ai/context/os]
18:23:30 ended    [context: github.com/gptscript-ai/context/os] [output=The local operating systems is Darwin, release 23.3.0]
18:23:30 sent     [main]
         content  [1] content | Waiting for model response...
         content  [1] content | <tool call> openFileNix -> {"file": "github.com/gptscript-ai/cli-demo", "port": "10993"}         content  [7] content | The local operating systems is Darwin, release 23.3.0
         content  [7] content | 

18:23:32 started  [open-file-nix(8)] [input={"file": "github.com/gptscript-ai/cli-demo", "port": "10993"}]
18:23:32 sent     [open-file-nix(8)]
18:23:32 ended    [open-file-nix(8)]
18:23:32 continue [main]
18:23:32 started  [context: github.com/gptscript-ai/context/os]
18:23:32 sent     [context: github.com/gptscript-ai/context/os]
18:23:32 ended    [context: github.com/gptscript-ai/context/os] [output=The local operating systems is Darwin, release 23.3.0]
18:23:32 sent     [main]
         content  [1] content | Waiting for model response... ○ Compiling /run ...
 ✓ Compiled /run in 1253ms (4996 modules)
 GET /run?file=github.com/gptscript-ai/cli-demo 200 in 1634ms

         content  [1] content | The task has been completed. The service was run, the port was obtained (10993), and the file "github.com/gptscript-ai/cli-demo" was opened using the openFileNix tool with the specified port.
         content  [9] content | The local operating systems is Darwin, release 23.3.0
         content  [9] content | 
18:23:35 ended    [main] [output=The task has been completed. The service was run, the port was obtained (10993), and the file \"githu...]

INPUT:

--file=github.com/gptscript-ai/cli-demo

OUTPUT:

The task has been completed. The service was run, the port was obtained (10993), and the file "github.com/gptscript-ai/cli-demo" was opened using the openFileNix tool with the specified port.
 POST /run?file=github.com/gptscript-ai/cli-demo 200 in 1633ms
 POST /run?file=github.com/gptscript-ai/cli-demo 200 in 11ms
 POST /run?file=github.com/gptscript-ai/cli-demo 200 in 5ms
 POST /run?file=github.com/gptscript-ai/cli-demo 200 in 9ms
 POST /run?file=github.com/gptscript-ai/cli-demo 200 in 7ms
 ⨯ unhandledRejection: error, status code: 401, message: Incorrect API key provided: asdf. You can find your API key at https://platform.openai.com/account/api-keys.
 ⨯ unhandledRejection: error, status code: 401, message: Incorrect API key provided: asdf. You can find your API key at https://platform.openai.com/account/api-keys.
thedadams commented 2 months ago

This should be fixed on main of gptscript.

sangee2004 commented 2 months ago

Teated with gptscript version - v0.0.0-dev-53f7fbde-dirty

This issue is not seen anymore.

When launching ui with anthropic provider , user is not prompted to enter openai api keys. Able to chat successfully with the provider using UI.