Enable Advanced Context Management for LLM Conversations

[DmitryHudrich]

User Story

1. As a user,
   I want Asya to store and reference conversation history even when LLM functions are activated,
   so that I can have a seamless and context-aware interaction with the assistant.

2. As a user,
   I want to export the current conversation context in JSON/XML format,
   so that I can review or edit it outside the application for further customization.

3. As a user,
   I want to upload an edited context back into the conversation in JSON/XML format,
   so that I can modify the flow or provide additional context for Asya to consider.

4. As a user,
   I want to clear the existing conversation context,
   so that I can reset the session and reduce token usage for more efficient interactions.

---

Current Behavior

• Asya does not provide users with options to export or modify conversation context.
• Context cannot be reset manually during a session.
• LLM responses may unnecessarily consume tokens when historical context is large.

---

Proposed Solution

1. Conversation History Storage:
   Retain conversation history when LLM is activated, as outlined in the previous request.

2. Export Context:

   • Provide a "Download Context" feature that exports the current conversation history in JSON or XML format.
   • Include metadata such as message timestamps, sender type (user/assistant), and any additional session details.

3. Upload Edited Context:

   • Allow users to upload a modified JSON/XML file containing the conversation context.
   • Validate the uploaded file for structure and content integrity.
   • Update the session with the new context, enabling users to shape the assistant's responses.

4. Clear Context:

   • Add a "Clear Context" button or command to reset the conversation history.
   • Notify users that clearing the context will start a fresh session and reduce token usage.

---

Use Case Scenarios

1. Scenario 1: Export Context

   • A user is working on a collaborative project with Asya.
   • They export the current conversation in JSON format to share with a teammate for further analysis.

2. Scenario 2: Edit and Upload Context

   • A user wants Asya to reference additional information not covered in the initial conversation.
   • They edit the exported context to include new data and upload the file back to Asya.
   • Asya uses the updated context to refine its responses.

3. Scenario 3: Clear Context

   • During a long session, the user notices that earlier parts of the conversation are no longer relevant.
   • They clear the context to start a new session, optimizing token usage and improving Asya's focus on the current topic.

---

Implementation Considerations

1. Export Functionality:

   • Use a REST API endpoint or in-app feature to generate downloadable JSON/XML files.
   • Example structure:

     {
       "session_id": "abc123",
       "messages": [
         { "sender": "user", "timestamp": "2025-01-07T12:00:00Z", "content": "Hello, Asya!" },
         { "sender": "asya", "timestamp": "2025-01-07T12:00:05Z", "content": "Hi! How can I assist you?" }
       ]
     }

2. Upload Functionality:

   • Create a secure mechanism to upload JSON/XML files via UI or API.
   • Validate the uploaded file to ensure it meets structural requirements.

3. Clear Context:

   • Implement a session reset function that removes stored context and notifies the user.

4. UI Enhancements:

   • Add buttons for exporting, uploading, and clearing context in the settings or conversation interface.
   • Provide feedback on actions (e.g., success, error messages).