GEMINI-CLI Settings Parameter Details

Introduction

#

Overview of the configuration and specific effects in GEMINI-CLI.

This also outlines the support scope of GEMINI-CLI parameters for CLI mode and A2A-server mode.

The parameters are referenced from specific sources. This article focuses on some important or confusing parameters, such as general.preferredEditor (in CLI mode), general.previewFeatures, etc., which will not be explained separately.

At the same time, this article highlights the gap between current A2A-server and CLI modes, referencing related features and PRs.

Gemini CLI configuration

general

#

checkpointing

#

Parameter Setting: general.checkpointing.enabled

Github PR: https://github.com/google-gemini/gemini-cli/pull/934

Supported in CLI/A2A-Server: Both supported

Function Description:

Supports checkpoint storage for conversation history. It only stores tool calls with awaiting_approval status in write operations like write, replace, etc.

This mechanism is designed to automatically save the current context and Git snapshot before the user approves a file modification operation (such as write or replace). This way, if the operation goes wrong, you can use this JSON file and Git hash to rollback or restore. Ordinary read-only operations (such as read_file) or operations that have already been executed will not trigger this save logic.

Checkpoints will be stored in the following directory:

${homedir}/.gemini/tmp/${projectPathHash}/checkpoints

Each tool call will be stored as a separate file, and the current git commit hash will be recorded. The specific file format is as follows:

${homedir}/.gemini/tmp/${projectPathHash}/checkpoints/${timestamp}-${tool.args.file_path}-${toolName}.json

enablePromptCompletion

#

Parameter Setting: general.enablePromptCompletion

Github PR: https://github.com/google-gemini/gemini-cli/pull/4691

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description: Uses the gemini-flash-2.5:nothinking model for real-time automatic prompt completion, similar to Search Suggestions provided by Google or Code Completion provided by GitHub Copilot.

sessionRetention

#

Parameter Setting:

• general.sessionRetention.enabled
• general.sessionRetention.maxAge
• general.sessionRetention.maxCount
• general.sessionRetention.minRetention

Github PR: https://github.com/google-gemini/gemini-cli/pull/7662 / https://github.com/google-gemini/gemini-cli/pull/4401

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Used to manage sessions stored on disk in CLI mode. By configuring maxAge and maxCount, you can set the maximum duration and maximum number of session records saved on disk, respectively. It also supports setting minRetention to set a minimum retention time to prevent maxAge or maxCount from being set too short.

privacy

#

usageStatisticsEnabled

#

Parameter Setting: privacy.usageStatisticsEnabled

Github PR: https://github.com/google-gemini/gemini-cli/pull/6217

Supported in CLI/A2A-Server: Only CLI supported (A2A-server enabled by default, not configurable)

Function Description:

Indicates whether the user allows Gemini CLI to collect anonymous usage data, similar to a privacy switch.

model

#

name

#

Parameter Setting: model.name

Supported in CLI/A2A-Server: Both supported (A2A-server supported via environment variables)

Function Description:

The model name of the main Agent, the default model name.

maxSessionTurns

#

Parameter Setting: model.maxSessionTurns

Github PR: https://github.com/google-gemini/gemini-cli/pull/3507

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

The maximum number of conversation turns in a session, designed to prevent infinite loops. Default is -1, which means unlimited.

summarizeToolOutput

#

Parameter Setting: model.summarizeToolOutput

Github PR: https://github.com/google-gemini/gemini-cli/pull/4140

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Used to control the length of tool output, summarizing and digesting when necessary.

Currently only supports summarization for the run_shell_command tool, using the gemini-2.5-flash-lite model configured in summarizer-shell in the packages/core/src/config/defaultModelConfigs.ts file.

This parameter can be configured as:

{
  "summarizeToolOutput": {
    "run_shell_command": {
      "tokenBudget": 2000
    }
  }
}

compressionThreshold

#

Parameter Setting: model.compressionThreshold

Github PR: https://github.com/google-gemini/gemini-cli/pull/12001 / https://github.com/google-gemini/gemini-cli/pull/12317

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

The threshold for triggering context compression, default is 0.5.

Assuming the model’s maximum input token is 128k, context compression will be automatically triggered when the input tokens (history conversation) reach 64k tokens by default.

The model used for context compression is also configured in chat-compression-* in the packages/core/src/config/defaultModelConfigs.ts file.

skipNextSpeakerCheck

#

Parameter Setting: model.skipNextSpeakerCheck

Github PR: https://github.com/google-gemini/gemini-cli/pull/5257

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Used to determine whether the model needs to proceed to the next step.

The model might occasionally end before execution is complete. To prevent such cases, there are several judgment rules:

1. If the last message is a tool call (function), a record The last message was a function response, so the model should speak next. will be added and execution will continue.
2. If the last message has no content, a record The last message was a filler model message with no content (nothing for user to act on), model should speak next. will be added and execution will continue.
3. Finally, the model corresponding to next-speaker-checker configured in the packages/core/src/config/defaultModelConfigs.ts file will be used to judge whether to continue execution. The default is gemini-2.5-flash-base.

modelConfigs (Important!!!)

#

aliases

#

Parameter Setting: modelConfigs.aliases

Github PR: https://github.com/google-gemini/gemini-cli/pull/12556

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

For default configurations, we can look at the packages/core/src/config/defaultModelConfigs.ts file.

The JSON is pasted below. For convenience, you can refer to the table:

{
  "base": {
    "modelConfig": {
      "generateContentConfig": {
        "temperature": 0,
        "topP": 1
      }
    }
  },
  "chat-base": {
    "extends": "base",
    "modelConfig": {
      "generateContentConfig": {
        "thinkingConfig": {
          "includeThoughts": true
        },
        "temperature": 1,
        "topP": 0.95,
        "topK": 64
      }
    }
  },
  "chat-base-2.5": {
    "extends": "chat-base",
    "modelConfig": {
      "generateContentConfig": {
        "thinkingConfig": {
          "thinkingBudget": 8192
        }
      }
    }
  },
  "chat-base-3": {
    "extends": "chat-base",
    "modelConfig": {
      "generateContentConfig": {
        "thinkingConfig": {
          "thinkingLevel": "HIGH"
        }
      }
    }
  },
  "gemini-3-pro-preview": {
    "extends": "chat-base-3",
    "modelConfig": {
      "model": "gemini-3-pro-preview"
    }
  },
  "gemini-2.5-pro": {
    "extends": "chat-base-2.5",
    "modelConfig": {
      "model": "gemini-2.5-pro"
    }
  },
  "gemini-2.5-flash": {
    "extends": "chat-base-2.5",
    "modelConfig": {
      "model": "gemini-2.5-flash"
    }
  },
  "gemini-2.5-flash-lite": {
    "extends": "chat-base-2.5",
    "modelConfig": {
      "model": "gemini-2.5-flash-lite"
    }
  },
  "gemini-2.5-flash-base": {
    "extends": "base",
    "modelConfig": {
      "model": "gemini-2.5-flash"
    }
  },
  "classifier": {
    "extends": "base",
    "modelConfig": {
      "model": "gemini-2.5-flash-lite",
      "generateContentConfig": {
        "maxOutputTokens": 1024,
        "thinkingConfig": {
          "thinkingBudget": 512
        }
      }
    }
  },
  "prompt-completion": {
    "extends": "base",
    "modelConfig": {
      "model": "gemini-2.5-flash-lite",
      "generateContentConfig": {
        "temperature": 0.3,
        "maxOutputTokens": 16000,
        "thinkingConfig": {
          "thinkingBudget": 0
        }
      }
    }
  },
  "edit-corrector": {
    "extends": "base",
    "modelConfig": {
      "model": "gemini-2.5-flash-lite",
      "generateContentConfig": {
        "thinkingConfig": {
          "thinkingBudget": 0
        }
      }
    }
  },
  "summarizer-default": {
    "extends": "base",
    "modelConfig": {
      "model": "gemini-2.5-flash-lite",
      "generateContentConfig": {
        "maxOutputTokens": 2000
      }
    }
  },
  "summarizer-shell": {
    "extends": "base",
    "modelConfig": {
      "model": "gemini-2.5-flash-lite",
      "generateContentConfig": {
        "maxOutputTokens": 2000
      }
    }
  },
  "web-search": {
    "extends": "gemini-2.5-flash-base",
    "modelConfig": {
      "generateContentConfig": {
        "tools": [
          {
            "googleSearch": {}
          }
        ]
      }
    }
  },
  "web-fetch": {
    "extends": "gemini-2.5-flash-base",
    "modelConfig": {
      "generateContentConfig": {
        "tools": [
          {
            "urlContext": {}
          }
        ]
      }
    }
  },
  "web-fetch-fallback": {
    "extends": "gemini-2.5-flash-base",
    "modelConfig": {}
  },
  "loop-detection": {
    "extends": "gemini-2.5-flash-base",
    "modelConfig": {}
  },
  "loop-detection-double-check": {
    "extends": "base",
    "modelConfig": {
      "model": "gemini-2.5-pro"
    }
  },
  "llm-edit-fixer": {
    "extends": "gemini-2.5-flash-base",
    "modelConfig": {}
  },
  "next-speaker-checker": {
    "extends": "gemini-2.5-flash-base",
    "modelConfig": {}
  },
  "chat-compression-3-pro": {
    "modelConfig": {
      "model": "gemini-3-pro-preview"
    }
  },
  "chat-compression-2.5-pro": {
    "modelConfig": {
      "model": "gemini-2.5-pro"
    }
  },
  "chat-compression-2.5-flash": {
    "modelConfig": {
      "model": "gemini-2.5-flash"
    }
  },
  "chat-compression-2.5-flash-lite": {
    "modelConfig": {
      "model": "gemini-2.5-flash-lite"
    }
  },
  "chat-compression-default": {
    "modelConfig": {
      "model": "gemini-2.5-pro"
    }
  }
}

overrides

#

Parameter Setting: modelConfigs.overrides

Github PR: https://github.com/google-gemini/gemini-cli/pull/12556

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Supports specific scenarios by directly overriding a model’s configuration, ignoring configurations in aliases, and forcing the use of overrides. The benefit is that when batch changes are needed, you only need to change it here.

When overrideScope is core, it applies to everything by default; otherwise, overrideScope needs to match the key above.

[
  {
    "match": {
      "model": "chat-base",
      "overrideScope": "core"
    },
    "modelConfig": {
      "generateContentConfig": {
        "temperature": 0.5
      }
    }
  },
  {
    "match": {
      "model": "chat-base",
      "overrideScope": "loop-detection"
    },
    "modelConfig": {
      "generateContentConfig": {
        "temperature": 0
      }
    }
  }
]

customAliases

#

Parameter Setting: modelConfigs.customAliases

Github PR: https://github.com/google-gemini/gemini-cli/pull/13546

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Supports custom aliases, but the priority is higher than aliases.

context

#

fileName

#

Parameter Setting: context.fileName

Github PR: https://github.com/google-gemini/gemini-cli/pull/1001

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Used to configure the file list for context memory.

includeDirectories

#

Parameter Setting: context.includeDirectories

Github PR: https://github.com/google-gemini/gemini-cli/pull/5354

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Directories other than the workspace that the model can see, which can be understood as supporting multiple workspaces.

loadMemoryFromIncludeDirectories

#

Parameter Setting: context.includeDirectories

Github PR: https://github.com/google-gemini/gemini-cli/pull/5354

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Supports loading the GEMINI.md file in .gemini from different workspace directories.

fileFiltering

#

Parameter Setting:

• context.fileFiltering.respectGitIgnore
• context.fileFiltering.respectGeminiIgnore
• context.fileFiltering.enableRecursiveFileSearch
• context.fileFiltering.disableFuzzySearch

Github PR: https://github.com/google-gemini/gemini-cli/pull/1290 / https://github.com/google-gemini/gemini-cli/pull/3727 / https://github.com/google-gemini/gemini-cli/pull/6087 /

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Whether to filter .gitignore and .geminiignore during file search, whether to support recursive retrieval, and whether to disable fuzzy retrieval.

tools

#

sandbox

#

Parameter Setting: tools.sandbox

Github PR: https://github.com/google-gemini/gemini-cli/pull/1154

Supported in CLI/A2A-Server: Only CLI supported, A2A-server unavailable

Function Description:

Whether to enable the sandbox or the sandbox image.

shell

#

Parameter Setting:

• tools.shell.enableInteractiveShell
• tools.shell.pager (ignored)
• tools.shell.showColor (ignored)
• tools.shell.inactivityTimeout

Github PR: https://github.com/google-gemini/gemini-cli/pull/10661 / https://github.com/google-gemini/gemini-cli/pull/13531

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

enableInteractiveShell determines which shell mode to use. If enabled, it uses node-pty; if not enabled, it uses child_process.spawn or child_process.exec.

inactivityTimeout is the maximum time for no output from a shell command, default is 5 minutes.

autoAccept

#

Parameter Setting: tools.autoAccept

Github PR: https://github.com/google-gemini/gemini-cli/pull/6665

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported (Official notes state server-side agent might not need sandbox association)

Function Description:

Auto-accept, divided into three modes: Default, *AUTO_EDIT, YOLO.

A2A-server currently does not support configuring *AUTO_EDIT.

core

#

Parameter Setting: tools.core

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Used to limit built-in tools.

allowed

#

Parameter Setting: tools.allowed

Github PR: https://github.com/google-gemini/gemini-cli/pull/6453

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Allows a set of tools to skip user confirmation. For shell commands, you can finely control whether to confirm specific commands, e.g., [“run_shell_command(git)”, “run_shell_command(npm test)”].

exclude

#

Parameter Setting: tools.exclude

Github PR: https://github.com/google-gemini/gemini-cli/pull/12728

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Used to exclude specific tools.

enableToolOutputTruncation

#

Parameter Setting: tools.enableToolOutputTruncation

Github PR: https://github.com/google-gemini/gemini-cli/pull/8039

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Truncates tool output based on character count. The truncation threshold is the minimum of the truncateToolOutputThreshold parameter and the remaining context window (model’s token limit * 4 minus consumed tokens).

Math.min(
      // Estimate remaining context window in characters (1 token ~= 4 chars).
      4 *
        (tokenLimit(this.model) - uiTelemetryService.getLastPromptTokenCount()),
      this.truncateToolOutputThreshold,
    )

truncateToolOutputThreshold

#

Parameter Setting: tools.truncateToolOutputThreshold

Github PR: https://github.com/google-gemini/gemini-cli/pull/8039

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Threshold for truncating tool output characters.

truncateToolOutputLines

#

Parameter Setting: tools.truncateToolOutputLines

Github PR: https://github.com/google-gemini/gemini-cli/pull/8039

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Threshold for truncating tool output lines.

enableMessageBusIntegration

#

Parameter Setting: tools.enableMessageBusIntegration

Github PR: https://github.com/google-gemini/gemini-cli/issues/7231

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Whether to enable message bus integration. This MessageBus has roughly the following functions:

• It serves as the primary context for tool execution. During execution, tools have access to the message bus in addition to their input parameters, making it the ideal place to store or retrieve contextual information.
• The policy engine is integrated into the message bus, enabling integration of some policy rules.
  • Such as whether to allow execution, deny execution, or ask the user.
  • safety check rules.
  • hook check rules.

enableHooks

#

Parameter Setting: tools.enableHooks

Github PR: https://github.com/google-gemini/gemini-cli/issues/9097

Supported in CLI/A2A-Server: Only CLI supported, hook registration and execution only supported in CLI

Function Description:

Allows hooks to intercept and interact with messages on the bus.

security

#

blockGitExtensions

#

Parameter Setting: security.blockGitExtensions

Github PR: https://github.com/google-gemini/gemini-cli/pull/12838

Supported in CLI/A2A-Server: Only CLI supported, A2A-server not supported

Function Description:

Blocks installing and loading extensions from git and github-release. Default supports four types: git, github-release, local, and link. Extensions here refer to gemini-cli extensions in ~/.gemini/extensions, which can be Mcp, hooks, memory context, etc.

Postscript

#

There are other parameters like hooks, advanced, mcpServers, telemetry, etc., which I won’t go into detail here. If interested, you can check the official Gemini CLI documentation directly.

Gemini CLI configuration