Configure options below — the preview updates in real time.

Root-Level Options

Top-level configuration including policy paths.

Additional security or operational policy files/directories to load.

Admin-enforced policy files (Enterprise override).

Use Write TodosEnable the write_todos tool.

Privacy

Privacy and usage statistics settings.

Usage StatisticsEnable collection of usage statistics.

General Settings

Editor preferences, update behavior, approval mode, and session settings.

The preferred editor executable to open files in (e.g., "code", "vim").

Vim ModeEnable Vim keybindings inside the CLI terminal.
Enable Auto UpdateEnable automatic background updates.
Enable Auto Update NotificationEnable update notification prompts.

Default approval mode for tool execution. "default" prompts user; "auto_edit" auto-approves code edits; "plan" is read-only.

DevToolsEnable DevTools inspector on launch.
Enable NotificationsEnable run-event notifications for action-required prompts and session completion (macOS only).
Retry Fetch ErrorsRetry requests on "fetch failed" network errors.

Maximum number of network retry attempts for requests to the main chat model.

Debug Keystroke LoggingEnable debug logging of raw keystrokes to the console.
Session RetentionEnable automatic session history cleanup and retention limits.

Auto-delete chats older than this period (e.g. "30d").

Minimum retention period safety limit (e.g. "1d").

CheckpointingEnable session checkpointing for rapid rollback and rewind recovery.

Plan Mode Settings

Configure planning artifacts directory and model routing.

The directory path where intermediate planning artifacts are stored.

Plan Model RoutingAutomatically switch between Pro (planning phase) and Flash (implementation phase) models.

Output

Output format settings.

The output format for CLI responses.

Model & Inference

Configure the Gemini model, context history, and loop prevention.

The Gemini LLM to use for conversations.

Max user/model/tool turns to keep in a session context window. -1 is unlimited.

The fraction of context capacity at which to trigger context compression (e.g., 0.5 = 50%).

Disable Loop DetectionDisable automatic detection and stopping of infinite agentic tool loops.
Skip Next Speaker CheckSkips validating if the AI should yield context.

Context & Search

Configure context file names, directories, and search behavior.

Target file names containing hierarchical instructions/memory.

Additional workspace context directories.

Include Directory TreeInclude directory tree in initial request.

Maximum directories to search for memory.

Load Memory From Include DirectoriesScan include directories for GEMINI.md files during memory refresh.
Respect .gitignoreRespect .gitignore files when searching.
Respect .geminiignoreRespect .geminiignore files when searching.
Enable Recursive File SearchEnable recursive @ reference completion.
Enable Fuzzy SearchEnable fuzzy matching when searching files in prompts.

Paths to additional ignore files beyond .gitignore (earlier takes precedence).

UI & Display

Control visual output including thinking display, warnings, footer elements, and accessibility.

The color theme for the CLI UI.

Auto Theme SwitchingSwitch themes automatically based on terminal background.

Controls whether the "chain of thought" output is visible inline.

Show Home Directory WarningWarn if launching Gemini CLI from the global home directory.
Show Compatibility WarningsShow warnings about terminal emulator or OS limits.
Show Shortcuts HintDisplay the "? for shortcuts" UI prompt above input.
Hide Context SummaryHide the active GEMINI.md / Context / MCP servers tag bar above input.
Hide Window TitleHide window title bar.
Hide TipsHide helpful tips in UI.
Hide BannerHide application banner.
Hide FooterHide footer from UI entirely.
Hide CWD in FooterRemove the Current Working Directory path from the footer.
Hide Sandbox Status in FooterRemove the container isolation status from the footer.
Hide Model Info in FooterRemove the model name and context usage from the footer.
Hide Context Usage PercentageHide the context window usage percentage indicator.
Show Footer LabelsDisplay descriptive header labels in the footer.
Show Memory UsageDisplay memory usage information.
Show Line NumbersShow line numbers in code blocks in chat.
Show CitationsShow citations for generated text.
Show Model Info in ChatShow the model name in chat for each AI turn.
Show User IdentityShow signed-in user identity.
Escape Pasted @ SymbolsEscape @ symbols in pasted text.
Use Alternate BufferUse alternate screen buffer.
Use Background ColorUse background colors in UI.
Incremental RenderingEnable incremental rendering (requires alternate buffer).
Show SpinnerShow spinner during operations.

Loading display mode.

Error display verbosity level.

Screen ReaderRender as plain-text for screen readers.

Tools

Tool execution, sandboxing, shell, and discovery settings.

Legacy full-process sandbox execution environment.

Enable Interactive ShellUse node-pty for interactive shell.

Pager command for shell output.

Shell Show ColorShow color in shell output.

Max seconds without shell output before timeout.

Shell Output EfficiencyEnable output efficiency optimizations.
Use RipgrepUse ripgrep for file content search.

Max characters for truncated tool outputs (0 disables).

Disable LLM CorrectionDisable LLM error correction for edits.

Security

Authentication, YOLO mode, folder trust, environment variable redaction, and extension security settings.

The currently selected authentication type.

Required authentication type (enterprise).

Tool SandboxingExperimental tool-level sandboxing.
Disable YOLO ModeForcibly block YOLO (auto-execute all tools) mode, even if triggered via CLI flags.
Enable Permanent Tool ApprovalEnable "Allow for all future sessions" option in tool confirmation dialogs.
Auto Add to Policy by DefaultMake "Allow for all future sessions" the default option in trusted workspaces.
Folder TrustEnable folder trust tracking.
Environment Variable RedactionEnable environment variable secret redaction.

Environment variables to always allow (bypass redaction).

Environment variables to always redact.

Enable ConsecaContext-aware security checker.
Block Git ExtensionsBlock installing extensions dynamically from untrusted remote Git repositories.

Regex pattern list. If populated, only extensions matching these patterns are executed.

Experimental & Advanced

Experimental features for tool output masking, plan mode, clipboard handling, and agents.

Tool Output MaskingEnable masking of verbose, redundant tool outputs to save context tokens.
Tool Output Masking: Protect Latest TurnNever mask the absolute latest turn.
Use OSC 52 for PastingRoute paste operations over ANSI OSC 52 (improves stability in SSH/tmux).
Use OSC 52 for CopyingRoute copy operations over ANSI OSC 52.
Plan Mode (Experimental)Enable Plan Mode.
Model SteeringEnable invisible hinting to guide the agent toward correct tool pipelines.
Direct Web FetchDownload webpage bodies fully into context rather than asking an LLM to pre-summarize them.
Enable AgentsEnable local/remote subagents.
Extension ManagementEnable extension management.
Extension ConfigEnable extension settings requests.
Extension RegistryEnable registry explore UI.
Just-In-Time ContextEnable Just-In-Time context loading.
Task TrackerEnable task tracker tools.

Billing

Billing and overage settings.

Strategy for handling billing overages.

IDE Integration

IDE integration settings.

IDE EnabledEnable IDE integration mode.

Advanced

Advanced settings for memory, DNS, and environment variables.

Auto Configure MemoryAuto-configure Node.js memory limits.

DNS resolution order.

Environment variables to exclude from context.

DEBUGDEBUG_MODE

Agents

Browser agent and agent override settings.

Browser HeadlessRun browser in headless mode.

Browser session mode.

Disable User Input (Browser)Disable user input during browser automation.

Skills

Agent skills settings.

Skills EnabledEnable Agent Skills.

List of disabled skill names.

Hooks Configuration

Hooks system settings.

Hooks EnabledCanonical toggle for the hooks system.
Hook NotificationsShow hook execution indicators.

MCP Servers

Model Context Protocol server definitions for external agent tools.

Allowlist of MCP servers. Only servers named here will be permitted to start.

Denylist of MCP servers to exclude.

Telemetry

Telemetry and observability settings.

Telemetry EnabledWhether or not telemetry is enabled.

Telemetry destination.

OTLP Exporter endpoint URL.

OTLP protocol.

Log PromptsInclude user prompt content in telemetry logs.

File path for local telemetry output.

Use CollectorUse external OTLP collector.