fix settings persistence

trying to fix HF Space

Gradio 6 port

gpt-image-1.5

gpt-4o-transcribe-diarize support

add gpt-5-pro, clean up model list

add o3-pro, remove gpt-4

fix recovery from previous errors

ignore mcp registry

require latest openai

ignore .env.sh

add explicit reasoning effort & verbosity controls

feat: GPT-5 in die Modellauswahl der Anwendung integrieren

basic gpt-5 support

image generateration 2nd generation

feed encrypted reasoning traces back to model

load .env properly

ignore locally installed JS artifacts

local MCP support

o3-low support

prevent scrolling issues when the textarea gets full

gradio update

move MCP approval handling

fix MCP call approval

Add optional UnrestrictedPython mode

Rename unrestricted mode flag

RestrictedPython hardening

Improve sandbox getattr and write guards

don't pass MCP results back to the LLM, as the Responses API does that internally. (Fixed invalid request exception).

Autofill API key from env

Prefill API key from env

Improve tool call presentation

Fix response format

Handle MCP progress events and fix response format

Preserve tool messages

Refine ChatMessage output while keeping streaming

Fix history import for gradio 5

Handle new gradio component serialization

Handle Gradio components in history

Refactor history import without temp files

Fix history import for images

Refine history parsing

Handle file dictionaries in history

Handle image imports

Fix history import with gradio 5

add basic Remote MCP support

Remove modal UI for MCP approvals

Switch approval modal to Gradio components

Use modal dialogs for MCP approvals

Handle MCP approval requests