Google has introduced Plan Mode in Gemini CLI, providing a read-only environment to review and plan code changes without the risk of altering any files.
Scientists confirm: This is the most effective way to get your cat’s attention, according to new research
Elderly Couple Refuses Reserved Seats—Viral Train Standoff Sparks Fiery Debate on Courtesy
Since its release in June 2025, Gemini CLI has continually expanded. Following integration with GitHub and the introduction of third-party extensions, as well as the upgrade to Gemini 3 Pro, Google’s command line tool has now been enhanced once again. On Wednesday, March 11, 2026, Google announced the launch of Plan Mode, a read-only planning mode designed to secure the planning phase.
A Planning Mode That Distinguishes Between Analysis and Execution
Plan Mode sets Gemini CLI into a restricted, read-only state. Google explains that in this mode, “the agent can browse your source code, search for patterns, and consult documentation”, as well as map dependencies, but it cannot modify any project files. Only its own internal plans, stored as Markdown files, can be edited.
Users can instruct Gemini CLI to “explore how to migrate this database” or “plan a new feature”, and it will identify dependencies and suggest a solution without the risk of premature code alterations.
To facilitate this analysis phase, the agent uses a set of limited tools:
- File system reading tools (read_file, grep_search, glob),
- Specialized search sub-agents like codebase_investigator,
- Read-only MCP tools, which allow viewing GitHub issues, PostgreSQL schemas, or Google Docs documents without altering the codebase,
- Web searching through Google Search.
Why You Should Never Reheat These Foods in the Microwave – The Hidden Dangers Experts Warn About
I tried the top 5 guard dogs—here’s what makes these breeds the ultimate protectors
Another significant addition is the ask_user tool. It allows the agent to pause its analysis to ask targeted questions, possibly to clarify an objective, confirm an architectural choice, or query the user about the location of a configuration file. “This two-way communication ensures that the final plan aligns perfectly with your vision before you commit to implementing it”, Google notes.
For more complex projects, Google highlights Conductor, an extension that utilizes Plan Mode to orchestrate multi-step development workflows. Conductor performs preliminary checks in read-only mode, then uses ask_user to have the user confirm decisions at each milestone. The team is working on integrating Conductor directly into Gemini CLI.
Conductor in Plan Mode:
Gemini CLI: Google’s AI Agent for Terminal
A Default Mode in Gemini CLI, Driven by Gemini 3.1 Pro
The separation between planning and execution adds an extra layer of security to development workflows. In sensitive operations (such as database migrations, major refactorings, or adding cross-cutting features), the risk of unintended modifications is a recurring issue with AI code agents. Plan Mode addresses this by requiring structured reflection time before any action on the code.
Plan Mode is enabled by default for all users of Gemini CLI. There are three ways to access it:
- The /plan command in the interface,
- The Shift+Tab keyboard shortcut to toggle between approval modes (Default, Auto-Edit, Plan),
- A natural language instruction, for example, “start a plan for…”.
Regarding models, Gemini CLI uses automatic routing to engage Gemini 3.1 Pro during the planning phase to ensure more robust architectural decisions. Once the plan is approved, the tool switches to a faster Flash model for implementation. Users who prefer not to use this mode can disable it using the /settings command.
Similar Posts
- Gemini Now Transcribes Audio Files: Discover How It Works!
- Gemini CLI Now Integrates With Figma, Stripe, Shopify: Enhance Your Workflow!
- Google Unveils Gemini CLI: Free Open Source AI Tool for Developers
- Google Chooses Gemini 3 Flash as Default Model: What You Need to Know!
- Google’s AI Code Agent Jules: How to Test Its Cutting-Edge Features!
Jordan Park writes in-depth reviews and editorial opinion pieces for Touch Reviews. With a background in UI/UX design, Jordan offers a unique perspective on device usability and user experience across smartphones, tablets, and mobile software.