Parse AI decompilation response and extract function decompilations by address.

Extract variable names from declarations and add to tracking set.

Extract variable names from declarations and add to tracking set.

Auto-populate missing model configurations from OPENAI_MODEL if available.

Resolve a raw tag emitted by the LLM. Returns ``(normalised_tag_id, is_new_dynamic)`` where *is_new_dynamic* is ``True`` when the tag was reg

Orchestrate second-pass unknown resolution in batches of 40. *llm_call_fn* has signature ``(prompt) -> (tool_calls, tokens_used, response_text, m

Restore aire_ prefixed function names to IDA defaults using MCP.

Restore protected content, like hilite.html.

Restore protected content, like hilite.html.

Resume a partial/failed index on a background thread.

Generate AI decompilation for the current function using shared gathering approach.

Run custom prompt correction analysis on current function.

Run fast look analysis on current function.

Save AI decompilation for a function address.

Save AI decompilation for a function address.

Schedule a UI-related async task on the UI worker thread. This won't block the pipeline worker.

Set a callable that renders the current agent state for prompt inclusion.

Check if a function should be filtered out based on name.

Show error dialog about missing config.

Truncate a code block to ``max_tokens`` tokens by keeping a head and a tail slice separated by a clear marker comment. Returns ``(trunca

Handles the core logic of streaming from LLM and saving results.

Strip user comments and aire_ prefixed function names from the current function. Returns: (aire_functions_list, success)

Simply return clean pseudocode with address prefixes for lines that have addresses, and 'cannotComment' for lines without addresses.

Convert block to list of message dicts for LLM API.

Aggressively truncate to get under budget. Args: target_tokens: Target token count (uses max_tokens if None) Re

Undo annotations from a specific analysis entry using MCP.

Creates the custom viewer window.

Creates the custom viewer window.

Creates the custom viewer window.

Display AI decompilation code with proper C++ syntax highlighting.

Display AI decompilation code with enhanced C++ syntax highlighting.

Display AI deobfuscation code with enhanced C++ syntax highlighting.

Calculate token estimate and priority.

Ensure UI message updates always run on IDA/Qt main thread.

Atomic IDA update optimized for stability.

Convert ConversationBlock objects to LLM input format.

Call LLM to generate summary. Works with any LLM provider (OpenAI, OpenRouter, local models, etc.) Does not use response_form

Pretty-print JSON but collapse short arrays onto a single line. Python's ``json.dumps(indent=…)`` puts every list element on its own line. T

Convert raw message history to ConversationTurn objects.