temp model for image reading

[iceljc]

PR Type

Other

---

Description

• Replace dynamic model selection with hardcoded "gpt-5-mini"

• Remove LLM provider service dependency

• Simplify completion provider initialization

---

Diagram Walkthrough

flowchart LR
  A["Dynamic Model Selection"] --> B["Hardcoded Model"]
  C["LLM Provider Service"] --> D["Direct Provider String"]
  E["Multi-modal Model Query"] --> F["Simple Model Name"]

Loading

File Walkthrough

Relevant files

Configuration changes

ReadImageFn.cs Hardcode model selection in image reading                                src/Plugins/BotSharp.Plugin.FileHandler/Functions/ReadImageFn.cs Remove LLM provider service dependency Replace dynamic model selection with hardcoded "gpt-5-mini" Simplify completion provider initialization +3/-4      ReadPdfFn.cs Hardcode model selection in PDF reading                                    src/Plugins/BotSharp.Plugin.FileHandler/Functions/ReadPdfFn.cs Remove LLM provider service dependency Replace dynamic model selection with hardcoded "gpt-5-mini" Simplify completion provider initialization +3/-4

---

[qodo-merge-pro]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Hardcoded Model Replacing dynamic provider/model resolution with hardcoded values reduces flexibility and may break multi-modal/image-specific behavior previously provided by 'gpt-4o'. Validate that 'gpt-5-mini' supports required capabilities and that environment-specific configuration is not needed. var provider = "openai"; var model = "gpt-5-mini"; var completion = CompletionProvider.GetChatCompletion(_services, provider: provider, model: model); var response = await completion.GetChatCompletions(agent, dialogs); return response.Content; Provider Assumption The provider is now hardcoded to 'openai'. Confirm this aligns with deployment environments and that fallbacks or configuration overrides are not required. Consider reading from configuration to avoid vendor lock-in. var provider = "openai"; var model = "gpt-5-mini"; var completion = CompletionProvider.GetChatCompletion(_services, provider: provider, model: model); var response = await completion.GetChatCompletions(agent, dialogs); return response.Content; Error Handling Context With initialization moved inline, ensure exceptions from provider/model mismatches are handled with actionable logs, since service-based validation is removed. Verify logging includes provider/model details. var provider = "openai"; var model = "gpt-5-mini"; var completion = CompletionProvider.GetChatCompletion(_services, provider: provider, model: model); var response = await completion.GetChatCompletions(agent, dialogs); return response.Content; } catch (Exception ex)

[qodo-merge-pro]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Use configurable multimodal model Hardcoding a potentially non-multimodal model can break image analysis. Make the provider/model configurable and default to a known multimodal-capable model to prevent runtime failures. This preserves flexibility across environments. src/Plugins/BotSharp.Plugin.FileHandler/Functions/ReadImageFn.cs [101-103] -var provider = "openai"; -var model = "gpt-5-mini"; +var provider = Environment.GetEnvironmentVariable("LLM_PROVIDER") ?? "openai"; +var model = Environment.GetEnvironmentVariable("BOTSHARP_VISION_MODEL") ?? "gpt-4o"; var completion = CompletionProvider.GetChatCompletion(_services, provider: provider, model: model); Apply / Chat Suggestion importance[1-10]: 9 __ Why: The suggestion correctly identifies that hardcoding gpt-5-mini in ReadImageFn is a critical issue, as this function requires a multimodal model which the hardcoded model may not be, potentially breaking the core functionality.	High
	Default to configurable multimodal model A hardcoded model that may lack multimodal support risks failing on PDFs with embedded images. Make provider/model configurable and default to a proven multimodal model to ensure reliable analysis. This avoids failures in varied deployments. src/Plugins/BotSharp.Plugin.FileHandler/Functions/ReadPdfFn.cs [92-94] -var provider = "openai"; -var model = "gpt-5-mini"; +var provider = Environment.GetEnvironmentVariable("LLM_PROVIDER") ?? "openai"; +var model = Environment.GetEnvironmentVariable("BOTSHARP_VISION_MODEL") ?? "gpt-4o"; var completion = CompletionProvider.GetChatCompletion(_services, provider: provider, model: model); Apply / Chat Suggestion importance[1-10]: 9 __ Why: The suggestion correctly identifies a critical risk, as ReadPdfFn may need to process images within PDFs, requiring a multimodal model. Hardcoding gpt-5-mini removes the guarantee of multimodal capability, potentially breaking functionality.	High
More