fix: fix simple_chat Responses tool schema + model discovery fallback

[gyliu513]

What does this PR do?

• Root cause: examples/agents/simple_chat.py sent tools without the required type discriminator for the Responses API, causing a 400. Also, examples/agents/utils.py assumed the client model schema always exposes model_type, which is not true for some deployments, leading to AttributeError during model selection.
• Fix: Send web search tool using Responses-compliant schema ({"type": "web_search"}) and adjust logging to match AgentEventLogger output. Add resilient model discovery helpers to resolve model id/type across client schema variants and default to LLM when type is missing.
• Notes: Changes are limited to examples; no runtime API behavior changes.

(stack) gualiu@gualiu-mac llama-stack-apps % python -m examples.agents.simple_chat --host localhost --port 8321
INFO:httpx:HTTP Request: GET http://localhost:8321/v1/shields "HTTP/1.1 200 OK"
No available shields. Disabling safety.
INFO:httpx:HTTP Request: GET http://localhost:8321/v1/models "HTTP/1.1 200 OK"
Using model: ollama/llama3.2:3b
INFO:httpx:HTTP Request: POST http://localhost:8321/v1/conversations "HTTP/1.1 200 OK"
User> Hello
INFO:httpx:HTTP Request: POST http://localhost:8321/v1/responses "HTTP/1.1 200 OK"
🤔 {"name": "search", "parameters": {"query": "hello world"}}
User> Search web for which players played in the winning team of the NBA western conference semifinals of 2024
INFO:httpx:HTTP Request: POST http://localhost:8321/v1/responses "HTTP/1.1 200 OK"
🤔

🔧 Executing web_search (server-side)...
🤔 I was unable to find any specific information on the 2024 NBA Western Conference Semifinals playoff winners. However, I can suggest some options to help you find the answer:

You can try searching for the official NBA website or social media channels for the latest updates and news.

Another option is to check online sports websites such as ESPN, Sports Illustrated, or CBS Sports, which provide comprehensive coverage of the NBA playoffs.

Feature/Issue validation/testing/test plan

Please describe the tests that you ran to verify your changes and relevant result summary. Provide instructions so it can be reproduced.
Please also list any relevant details for your test configuration or test plan.

• Test A
  Logs for Test A

• Test B
  Logs for Test B

Sources

Please link relevant resources if necessary.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

Thanks for contributing 🎉!

[gyliu513]

@cdoern @leseb ^^

[raghotham]

it is not clear that any developer will write code like this when creating agents using llama stack client. Can you make it so that the code here is something a new developer can just copy? We dont need any backward compatibility here either. We could just use the latest version. We can have copies of the examples for older versions if needed.

[gyliu513]

Done

[gyliu513]

@raghotham can you help check if this can be merged? Thanks!

[raghotham]

do we have to run a chat completion to see if the model supports chat? we already have model type: https://github.com/llamastack/llama-stack/blob/ffa98595e696c7ab3e0e933d0ed75375ee1d7b84/src/llama_stack_api/models/models.py#L23

[gyliu513]

@raghotham I can see there are some models with llm type still do not support chat, like ollama/all-minilm:latest.

(llama-stack) (base) gualiu@gualiu-mac llama-stack % curl -s http://localhost:8321/v1/models \
  | jq '.data[] | select(.id=="ollama/all-minilm:latest")'
{
  "id": "ollama/all-minilm:latest",
  "object": "model",
  "created": 1769569923,
  "owned_by": "llama_stack",
  "custom_metadata": {
    "model_type": "llm",
    "provider_id": "ollama",
    "provider_resource_id": "all-minilm:latest"
  }
}

But this model do not support chat.

(stack) gualiu@gualiu-mac llama-stack-apps % python -m examples.agents.simple_chat --host localhost --port 8321  --model
_id ollama/all-minilm:latest
INFO:httpx:HTTP Request: GET http://localhost:8321/v1/models "HTTP/1.1 200 OK"
Using model: ollama/all-minilm:latest
INFO:httpx:HTTP Request: POST http://localhost:8321/v1/conversations "HTTP/1.1 200 OK"
User> Hello
INFO:httpx:HTTP Request: POST http://localhost:8321/v1/responses "HTTP/1.1 200 OK"
🤔
❌ Turn failed: Error code: 400 - {'error': {'message': '"all-minilm:latest" does not support chat', 'type': 'api_error', 'param': None, 'code': None}}
User> Search web for which players played in the winning team of the NBA western conference semifinals of 2024
INFO:httpx:HTTP Request: POST http://localhost:8321/v1/responses "HTTP/1.1 200 OK"
🤔
❌ Turn failed: Error code: 400 - {'error': {'message': '"all-minilm:latest" does not support chat', 'type': 'api_error', 'param': None, 'code': None}}

I think besides model_type, we may need to add a new field named as capability for the model, the capability can be chat, completion, tool_calling etc, comments?