Insight

🎯 Purpose

To analyze the user's active Antigravity session and conversation history to generate highly relevant, actionable recommendations for workflow optimization. This includes identifying repetitive instructions to add to AGENTS.md, suggesting specific Antigravity features (like MCP Servers or Hooks), and defining reusable usage patterns.

🚪 Gating & Trigger Conditions

• When to invoke: When the user explicitly requests an analysis of "this conversation", asks to review the current session's workflow, or runs the specific skill command.
• When NOT to invoke: During standard code generation, debugging, or when the conversation history is too short to extract meaningful patterns (e.g., at the very beginning of a session).

📥 Input Specifications

• conversation_history: The active context window containing the back-and-forth dialogue, prompts, and responses from the current session.
• features_reference: The static list of Antigravity capabilities (MCP Servers, Custom Skills, Hooks, Headless Mode, Task Agents) provided in the system instructions.

⚙️ Execution Instructions (Workflow)

1. Analyze Active Context: Scan the current conversation_history loaded in your memory.
2. Extract Repetitive Patterns: Identify specific instructions, corrections, or preferences the user has had to state multiple times during this session (e.g., "always use TypeScript", "remember to format the output").
3. Map to AGENTS.md: Draft additions for AGENTS.md based only on the repetitive patterns identified in Step 2.
4. Feature Matching: Review the features_reference list. Select 2-3 features that directly solve pain points or automate the repetitive tasks observed in the current conversation. Include 2-3 items for each category if applicable.
5. Formulate Usage Patterns: Identify 2-3 broader workflow habits from the session that could be optimized, drafting copyable prompts to help the user adopt these habits in future sessions.
6. Construct Output: Generate a well-structured Markdown document organizing the findings into clear sections.
7. Validation (Fallback): If the generated output contains conversational filler outside the requested Markdown structure, discard the filler and output only the strict Markdown report.

📤 Output Specifications

• Format: A formatted Markdown document.
• Structure Template:

  # 📊 Antigravity Session Analysis
  
  ## 📝 AGENTS.md Additions
  *(Based on repeated instructions in this conversation)*
  
  - **[Proposed Addition Text]**
    - **Why:** [1 sentence explaining why this helps based on the session]
    - **Implementation:** [Instructions for where to add this in AGENTS.md]
  
  ## 🚀 Features to Try
  
  ### [Feature Name]
  - **Overview:** [What it does in one line]
  - **Why it fits your workflow:** [Explanation based on the active conversation context]
  - **How to use it:**
    \`\`\`[language]
    [Actual command or config to copy]
    \`\`\`
  
  ## 🔄 Recommended Usage Patterns
  
  ### [Short Title]
  - **Suggestion:** [1-2 sentence summary]
  - **Details:** [3-4 sentences explaining how this applies to the work done in this session]
  - **Try this prompt:**
    > [A specific prompt to copy and try]
  
  ## 📁 References & Directory Structure
  - AGENTS.md: The target file for global agent instructions.
  - .agents/settings.json: Configuration file referenced for Hooks.
  - .agents/skills/: Directory referenced for Custom Skills.