docs: add Antigravity ide prompt

[cwdhf]

add Antigravity ide : Fast Prompt.txt、Planning Prompt.txt and Tools.json file

Summary by CodeRabbit

• New Features

  • Added nine tools to the system including browser operations, search capabilities, command tracking, file editing, image generation, and resource management.
  • Introduced structured task-management framework with planning, execution, and verification modes.

• Documentation

  • Added comprehensive agent guidelines and system configuration documentation.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Walkthrough

Three new declarative files added to the Antigravity directory: two comprehensive prompt/configuration text files (Fast Prompt.txt and Planning Prompt.txt) defining agent behavior, workflows, and task management semantics, and one JSON manifest (Tools.json) specifying nine tools with structured parameter schemas for browser automation, search, image generation, and file editing operations.

Changes

Cohort / File(s)	Change Summary
Agent Prompts Antigravity/Fast Prompt.txt, Antigravity/Planning Prompt.txt	Two new declarative prompt files: Fast Prompt.txt defines agent identity, tool usage guidelines, web development workflows, SEO practices, and communication style; Planning Prompt.txt introduces an AGENTIC mode with task boundaries, artifact templates (task.md, implementation_plan.md, walkthrough.md), mode state machine (PLANNING, EXECUTION, VERIFICATION), and workflow management conventions.
Tools Manifest Antigravity/Tools.json	New JSON schema manifest cataloging nine tools (browser_subagent, codebase_search, command_status, find_by_name, generate_image, grep_search, list_dir, list_resources, multi_replace_file_content) with detailed parameter definitions, required field constraints, and metadata for each tool's usage and behavior.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~35 minutes

• Tools.json parameter schema completeness: Verify all required fields are properly specified, especially the complex multi_replace_file_content structure with nested ReplacementChunks and metadata.
• Prompt semantic consistency: Ensure Fast Prompt.txt and Planning Prompt.txt align on tool naming, workflow conventions, and task artifact formatting (both reference tools and workflows that should be mutually compatible).
• Tool interface contract: Check that parameter schemas in Tools.json match the expected tool invocation patterns described in the prompt files.

Possibly related PRs

• Add planning mode system prompt #305: Overlapping planning-mode specification with task_boundary and notify_user tool definitions that align with the new Planning Prompt.txt agentic mode semantics.

Poem

🐰 Behold, the agent's new instruction tome,
Fast prompts and planning workflows find their home,
Nine tools now catalogued in JSON's neat array,
Where browser, search, and file-edit come to play.
The Antigravity mind takes deeper form today! ✨

Pre-merge checks and finishing touches

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: adding Antigravity IDE-related prompt documentation. It is concise, specific, and clearly indicates this is a documentation addition.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

📜 Recent review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between b77ac4d and 44a8603.

📒 Files selected for processing (3)

• Antigravity/Fast Prompt.txt (1 hunks)
• Antigravity/Planning Prompt.txt (1 hunks)
• Antigravity/Tools.json (1 hunks)

🧰 Additional context used

🧠 Learnings (2)

📓 Common learnings

Learnt from: thedtvn
Repo: x1xhlol/system-prompts-and-models-of-ai-tools PR: 307
File: Grok/Ani.md:35-35
Timestamp: 2025-11-20T11:10:16.064Z
Learning: In the repository x1xhlol/system-prompts-and-models-of-ai-tools, files contain system prompts extracted via reverse engineering or MITM attacks from AI tools. These prompts should be preserved exactly as found, including any malformed tags, unusual formatting, or apparent errors, as they represent the authentic source material.

📚 Learning: 2025-11-20T11:10:16.064Z

Learnt from: thedtvn
Repo: x1xhlol/system-prompts-and-models-of-ai-tools PR: 307
File: Grok/Ani.md:35-35
Timestamp: 2025-11-20T11:10:16.064Z
Learning: In the repository x1xhlol/system-prompts-and-models-of-ai-tools, files contain system prompts extracted via reverse engineering or MITM attacks from AI tools. These prompts should be preserved exactly as found, including any malformed tags, unusual formatting, or apparent errors, as they represent the authentic source material.

Applied to files:

• Antigravity/Planning Prompt.txt
• Antigravity/Fast Prompt.txt

🪛 LanguageTool

Antigravity/Planning Prompt.txt

[style] ~75-~75: This phrase is redundant. Consider writing “just” or “simply”.
Context: ...S SUFFICIENT COMPLEXITY TO THE TASK. If just simply responding to the user in natural langu...

(JUST_RECENTLY)

Antigravity/Fast Prompt.txt

[grammar] ~27-~27: Use a hyphen to join words.
Context: ...ed, harmonious color palettes (e.g., HSL tailored colors, sleek dark modes). -...

(QB_NEW_EN_HYPHEN)

---

[grammar] ~32-~32: Use a hyphen to join words.
Context: ...ke a design that feels premium and state of the art. Avoid creating simple minimu...

(QB_NEW_EN_HYPHEN)

---

[grammar] ~32-~32: Use a hyphen to join words.
Context: ...a design that feels premium and state of the art. Avoid creating simple minimum v...

(QB_NEW_EN_HYPHEN)

---

[grammar] ~32-~32: Use a hyphen to join words.
Context: ...sign that feels premium and state of the art. Avoid creating simple minimum viabl...

(QB_NEW_EN_HYPHEN)

---

[grammar] ~69-~69: Ensure spelling is correct
Context: ...to pay attention to. Do not respond to nor acknowledge those messages, but do foll...

(QB_NEW_EN_ORTHOGRAPHY_ERROR_IDS_1)

---

[style] ~81-~81: The phrase ‘have the ability to’ might be wordy. Consider using “can”.
Context: ..._global]> </user_rules> You have the ability to use and create workflows, which are wel...

(HAS_THE_ABILITY_TO)

---

[uncategorized] ~82-~82: Did you mean the formatting language “Markdown” (= proper noun)?
Context: ...follow the following YAML frontmatter + markdown format: --- description: [short title, ...

(MARKDOWN_NNP)

🔇 Additional comments (7)

Antigravity/Fast Prompt.txt (2)

1-66: Overall structure and content are well-organized; authenticity of extracted prompt should be preserved.

The Fast Prompt file is well-structured with clear sections (identity, tool calling, web development, workflows, communication style) that define agent behavior comprehensively. However, per the repository's learnings about preserving authenticity of reverse-engineered prompts, any formatting or grammar variations should be retained as they represent the original source material.

---

71-79: Repository access unavailable — unable to verify file content.

The repository x1xhlol/system-prompts-and-models-of-ai-tools could not be cloned, preventing verification of whether the file Antigravity/Fast Prompt.txt exists at lines 71-79 or whether the content matches the review comment's claims.

Without access to the actual file:

• Cannot confirm the <user_rules> and <MEMORY[user_global]> sections exist as shown
• Cannot assess whether this represents authentic extracted material or a test case
• Cannot verify the intended purpose and context of this content
• Cannot determine if the safety concern raised is valid for this specific repository

The review comment's core concern about permissive content policy overrides cannot be validated or refuted without examining the actual file.

Antigravity/Tools.json (2)

1-50: JSON schema structure is sound with well-defined tools and parameter schemas.

The tools are consistently structured with clear descriptions and properly required parameters. The schema follows a logical pattern across all nine tools, making it maintainable and usable.

---

234-319: Verify complex multi_replace_file_content schema alignment with actual tool usage.

The multi_replace_file_content tool has the most complex schema with nested objects, arrays, and metadata. Ensure that:

1. The ReplacementChunks structure matches the actual tool's implementation expectations
2. The ArtifactMetadata field aligns with the Planning Prompt artifact types (implementation_plan, walkthrough, task, other)
3. All required fields in the nested structures are actually mandatory in tool invocation

Consider spot-checking against actual tool invocations to confirm this schema is both accurate and complete. The Complexity (1-10 rating), TargetLintErrorIds, and ArtifactMetadata fields may need validation that they're enforced consistently.

Antigravity/Planning Prompt.txt (3)

1-68: Agentic mode framework is well-structured and clearly documented.

The Planning Prompt establishes clear task boundaries, mode transitions (PLANNING → EXECUTION → VERIFICATION), and artifact management semantics. The structured approach to complex task decomposition (via task.md, implementation_plan.md, walkthrough.md) provides good scaffolding for multi-step coding work.

---

75-95: Clarify file completeness and address minor redundancy per authenticity principle.

Two concerns:

1. File truncation: The provided file appears to end abruptly at line 144 mid-sentence ("Update existing walkthrough"). Verify that the full file is included in the PR and not truncated.

2. Minor redundancy (line 75): LanguageTool flags "just simply" as redundant. Per the repository's principle of preserving authentic extracted prompts, clarify whether this text is from the original source or introduced by the PR contributor. If it's original, preserve it; if it's new, consider simplifying to either "just" or "simply".

---

78-88: Mode descriptions provide clear state machine for task progression.

The three modes (PLANNING, EXECUTION, VERIFICATION) with explicit backtracking rules (lines 85, 87) establish a principled workflow. The guidance to stay within the same TaskName when backtracking mid-task, but switch modes, is a useful pattern that prevents task proliferation.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}