feat(prompts): implement Topic-Action-Summary model for verbosity reduction

[Abhijit-2592]

Summary

This PR implements the Topic Model (refined from the experimental Topic & Update model) and introduces a new experimental setting, topicUpdateNarration, to control it. This model significantly reduces terminal noise and 'excessive scrolling' during multi-turn tasks by suppressing repetitive tool-call narration and focusing on broad logical phases.

Details

The current 'Explain Before Acting' mandate forced the model to narrate every tool call, creating massive scroll noise when combined with high-volume output. This refactor:

• Mandates a Topic Model: Every task or broad phase (e.g., Research, Implementation) begins with a single, high-signal header in the format Topic: <Phase> : <Summary>.
• Enforces Silence: All subsequent tool calls and reasoning within that Topic are executed in complete silence (zero text output).
• Streamlined Logic: Removes the experimental 'Update' heartbeat to further reduce verbosity, relying on broad logical transitions for signaling progress.
• Safety Preservation: Explicitly preserves mandatory safety explanations for critical shell commands, ensuring security rules override the silence mandate.

Configuration & Implementation

• Added experimental.topicUpdateNarration to the settings schema (packages/cli/src/config/settingsSchema.ts).
• Updated the core Config and loadCliConfig to support and map this new setting.
• Refactored PromptProvider and snippets.ts to dynamically adjust system instructions based on the setting.
• Fixed a syntax error in snippets.ts (escaped backtick) and corrected broken nesting in settingsSchema.ts.
• Regenerated settings JSON schema and documentation.

Related Issues

Fixes #21304 (Context Verbosity Epic - Phase 1)

How to Validate

1. Enable the setting: Add "topicUpdateNarration": true to your .gemini/settings.json under the experimental block.
2. Verify prompt injection: Run with DEBUG=gemini-cli:prompts or check config list to ensure the setting is active.
3. Run a multi-turn task:
   • Success Turn 1: The agent MUST start with a header like Topic: <Research> : Analyzing code patterns followed by tools.
   • Success Turn 2+: Subsequent turns within the same phase MUST contain only tool calls and zero text output.

Recommended Test Prompts

1. Bug Fix Flow: "Find why the unit tests in packages/core/src/routing are failing, fix the underlying issue, and verify the fix by running the tests again."
2. Architectural Audit: "Analyze the codebase to find all places where we use vi.stubEnv. Once found, pick one test file in packages/core and refactor it to use a custom test helper instead, then confirm the tests still pass."
3. Tool Implementation: "Research how built-in tools are registered in the core package. Then, plan and implement a new 'hello-world' tool that simply returns a greeting string."
4. Cross-Package Feature: "I want to add a new 'telemetry' setting to the experimental section. First, find where settings are defined in both packages/cli and packages/core. Then, plan the change and implement the new setting in both packages."
5. Dependency Cleanup: "Identify all unused imports in packages/core/src/prompts/promptProvider.ts. Remove them, and then check if the package still builds correctly."

Pre-Merge Checklist

• Updated relevant documentation and README (if needed)
• Added/updated tests (if needed)
• Noted breaking changes (if any)
• Validated on required platforms/methods:
  • MacOS
    • npm run

[gemini-code-assist]

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces a "Topic-Action-Summary" communication model within the system instructions to significantly reduce verbose output and improve clarity during multi-turn tasks. The change aims to streamline the interaction by mandating a concise plan summary at the start of each phase, followed by silent tool execution, and clear signaling of strategic pivots, ultimately addressing the problem of excessive scrolling and enhancing user experience.

Highlights

• Topic-Action-Summary Model: Implemented a new communication pattern where tasks begin with a high-signal, one-line summary of the proposed plan, followed by silent execution of tool calls.
• Enforced Action Silence: Mandated that all subsequent responses within the same topic must contain only tool calls with zero text output, reducing terminal noise.
• Strategic Transition Signaling: Introduced a mechanism for signaling major pivots or transitions to new broad topics with a single high-signal line, maintaining transparency during strategy shifts.
• Refined Tone and Style Guidelines: Updated operational guidelines to emphasize extreme brevity, high-signal output, and the importance of absolute silence during action phases, aligning with the new communication model.

Changelog

• packages/core/src/prompts/snippets.ts
  • Replaced the previous 'Explain Before Acting' mandate with a detailed 'Topic-Action-Summary' communication pattern, outlining specific rules for initial topic statements, silent action phases, strategic transitions, and final summaries.
  • Modified tone and style guidelines to explicitly require high-signal output, extreme brevity (aiming for one line of text output per response), and absolute silence during action phases, reinforcing the new communication model.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

The pull request introduces a new "Topic-Action-Summary" communication model. However, a critical vulnerability has been identified: the "absolute silence" mandate for the Action phase directly conflicts with the mandatory safety rule requiring explanations for critical shell commands. This weakens a key safety control, potentially leading to unpredictable model behavior and compromising user understanding of dangerous commands. An explicit exception for safety-critical explanations from the silence mandate is recommended to resolve this conflict, aligning with the principle of explicit safety instructions for agent deviations.

[github-actions]

Size Change: +3.46 kB (+0.01%)

Total Size: 26.1 MB

Filename	Size	Change
./bundle/chunk-IX7UT4XO.js	0 B	-13.4 MB (removed)	🏆
./bundle/chunk-S363URX6.js	0 B	-3.62 MB (removed)	🏆
./bundle/core-BITOVHVD.js	0 B	-40.3 kB (removed)	🏆
./bundle/devtoolsService-XNHSIYXL.js	0 B	-27.7 kB (removed)	🏆
./bundle/interactiveCli-GJYFQR3H.js	0 B	-1.59 MB (removed)	🏆
./bundle/oauth2-provider-QE7XPRTD.js	0 B	-9.19 kB (removed)	🏆
./bundle/chunk-25R3AHNP.js	13.4 MB	+13.4 MB (new file)	🆕
./bundle/chunk-ZUMB2CSN.js	3.62 MB	+3.62 MB (new file)	🆕
./bundle/core-CTVQ7KWK.js	40.3 kB	+40.3 kB (new file)	🆕
./bundle/devtoolsService-CP4O35KV.js	27.7 kB	+27.7 kB (new file)	🆕
./bundle/interactiveCli-K4CCP3PM.js	1.59 MB	+1.59 MB (new file)	🆕
./bundle/oauth2-provider-JQF7RVJK.js	9.19 kB	+9.19 kB (new file)	🆕

ℹ️ View Unchanged

Filename	Size	Change
./bundle/chunk-34MYV7JD.js	2.45 kB	0 B
./bundle/chunk-37ZTTFQF.js	966 kB	0 B
./bundle/chunk-5AUYMPVF.js	858 B	0 B
./bundle/chunk-664ZODQF.js	124 kB	0 B
./bundle/chunk-BXZA3XKQ.js	1.95 MB	0 B
./bundle/chunk-DAHVX5MI.js	206 kB	0 B
./bundle/chunk-IUUIT4SU.js	56.5 kB	0 B
./bundle/chunk-RJTRUG2J.js	39.8 kB	0 B
./bundle/devtools-36NN55EP.js	696 kB	0 B
./bundle/dist-T73EYRDX.js	356 B	0 B
./bundle/gemini.js	695 kB	+71 B (+0.01%)
./bundle/getMachineId-bsd-TXG52NKR.js	1.55 kB	0 B
./bundle/getMachineId-darwin-7OE4DDZ6.js	1.55 kB	0 B
./bundle/getMachineId-linux-SHIFKOOX.js	1.34 kB	0 B
./bundle/getMachineId-unsupported-5U5DOEYY.js	1.06 kB	0 B
./bundle/getMachineId-win-6KLLGOI4.js	1.72 kB	0 B
./bundle/memoryDiscovery-ZBBGU3A2.js	922 B	0 B
./bundle/multipart-parser-KPBZEGQU.js	11.7 kB	0 B
./bundle/node_modules/@google/gemini-cli-devtools/dist/client/main.js	221 kB	0 B
./bundle/node_modules/@google/gemini-cli-devtools/dist/src/_client-assets.js	227 kB	0 B
./bundle/node_modules/@google/gemini-cli-devtools/dist/src/index.js	11.5 kB	0 B
./bundle/node_modules/@google/gemini-cli-devtools/dist/src/types.js	132 B	0 B
./bundle/sandbox-macos-permissive-open.sb	890 B	0 B
./bundle/sandbox-macos-permissive-proxied.sb	1.31 kB	0 B
./bundle/sandbox-macos-restrictive-open.sb	3.36 kB	0 B
./bundle/sandbox-macos-restrictive-proxied.sb	3.56 kB	0 B
./bundle/sandbox-macos-strict-open.sb	4.82 kB	0 B
./bundle/sandbox-macos-strict-proxied.sb	5.02 kB	0 B
./bundle/src-QVCVGIUX.js	47 kB	0 B
./bundle/tree-sitter-7U6MW5PS.js	274 kB	0 B
./bundle/tree-sitter-bash-34ZGLXVX.js	1.84 MB	0 B
./bundle/undici-4X2YZID5.js	360 B	0 B

_{compressed-size-action}