- Updated default model configurations (Gemini 3/3.1) to enforce `includeThoughts: true` and `ThinkingLevel.HIGH` and updated schema, doc and tests to reflect this.
- Extended the code assist converters to safely extract and format the `thought` field, preventing type errors during `CountToken` API calls.
- Implemented real-time `extractPlanEntries` parsing using regex that gracefully extracts markdown task lists (e.g., `[TODO]`, `[x]`, `[]`) while safely ignoring content inside code blocks.
- Added logic in the ACP client that incrementally reads the agent's thought buffer and emits `sessionUpdate: 'plan'` events directly to the IDE whenever the tracked tasks change.
- Formatted `agent_thought_chunk` outputs to replace escaped literal newlines (`\n`) with actual native newlines, ensuring the IDE chat displays main thought sections cleanly without visual artifacts.
- Added comprehensive unit tests for the plan parser and ACP integration.
CCPA uses a different format than GenAi. This adds conversion code to get it to the right format.
Note that this doesn't work against the current ccpa staging server, The changes it needs are in cl/770266927
Sri Pasumarthi