v0.43.0
Gemini CLISummary
This release introduces significant new capabilities, including the ability to export and import sessions, and the addition of Local and Remote Subagent Protocols. Numerous bug fixes address stability, including resolving chat corruption, preventing hangs during OAuth, and improving tool execution reliability. Users will also benefit from clearer documentation regarding Auto Memory and improved CLI UX in various scenarios.
New Features
- Added support for exporting the current session to a file and importing it via a flag.
- Added Machine Hostname to the CLI interface display.
- Introduced LocalSubagentProtocol behind the AgentProtocol interface.
- Added RemoteSubagentProtocol behind the AgentProtocol interface.
- Enabled the model to use the edit tool specifically for surgical edits.
- Added shell command safety evaluations.
- Prefix tool call IDs with tool names to support tool rendering in ACP compliant IDEs.
Bug Fixes
- Fixed a typo related to model steering for surgical edits.
- Rejected numeric project IDs when setting GOOGLE_CLOUD_PROJECT.
- Fixed a chat corruption bug within the context manager.
- Provided JSON output for AgentExecutionStopped events in non-interactive mode.
- Fixed handling of invalid custom plans directories, preventing crashes.
- Resolved a race condition when waiting for tool completion in the a2a-server.
- Randomized sandbox container names to prevent conflicts.
- Fixed hysteresis issues in asynchronous context management pipelines.
- Preserved the system PATH environment variable in Git operations to fix ENOENT errors.
- Fixed an issue where transcribed text would not display after releasing the spacebar.
- Prevented false command conflicts when launching the tool from the home directory.
- Fixed an error where a function response turn did not immediately follow a function call.
- Resolved parallel tool call streaming ID collisions.
- Fixed treating HTTP 404 responses as 405 in StreamableHTTPClientTransport for MCP.
- Prevented duplicate SessionStart systemMessage rendering.
- Prevented an infinite thought loop in ACP mode by disabling nextSpeakerCheck.
- Used a static tool name in the confirmation prompt to avoid parsing errors.
- Fixed tool turn handling in NumericalClassifierStrategy to prevent 400 Bad Request errors.
- Handled malformed projects.json files gracefully in ProjectRegistry.
- Fixed an issue where the tool explanation was incorrectly placed in the thought stream instead of tool call content.
- Fixed an issue preventing redirection in YOLO and AUTO_EDIT modes without sandboxing.
- Fixed an issue preventing silent hangs during OAuth authentication on headless Linux systems.
- Fixed an issue where read-only settings scopes were visible in the CLI.
- Fixed an issue where the /memory add subcommand was visible when memoryV2 is enabled.
- Fixed an issue where tool responses were dropped without an explicit error being thrown.
- Fixed an issue with maintainer identification in the lifecycle manager.
- Fixed an issue where quotes were missing around the session ID in the resume tip UI.
- Fixed an issue with the context snapshotter model configuration.
- Fixed an issue preventing extension installation from SSH repositories.
Improvements
- Clarified that Auto Memory proposes memory updates and skills.
- Improved UX for listing MCPs in untrusted folders.
- Migrated core tools to use the native ToolDisplay property, fixing UI rendering.
- Removed unnecessary wrapping of arguments.
- Hid read-only settings scopes from the CLI.
- Removed noisy theme registration logs from the terminal.
- Implemented codebase-aware effort level triage in CI.
- Improved the snapshotter functionality for context management.
- Added gutter width calculation to the input prompt width calculation.