Managing Stack Context

Managing Stack Context Skill

Purpose

The Antigravity factory uses a 7-Phase Meta-Orchestration loop that is stack-agnostic. However, Phase 4 (Build) and Phase 5 (Test) heavily depend on knowing what language/framework the project uses (e.g., Python vs C#).

This skill provides explicit tooling to query and definitively set the active SDLC_Stack.

When to Use

Use this skill at the beginning of a project session or when switching between different components of a polyglot repository to ensure the AI agent operates with the correct environment context and workflow constraints.

Prerequisites

• Conda environment cursor-factory must be active.
• Access to .agent/config/stack-configurations.json.

Process

To manage stack context effectively, follow these steps. First, perform a comprehensive repository audit to identify key indicator files (e.g., requirements.txt, package.json). Second, query the existing execution context using the get_active_stack tool to detect any state drift. Third, implement stack realignment by running the set_active_stack command with the appropriate stack key. Finally, verify and persist the state by confirming the update with get_active_stack and ensuring the Memory MCP graph is synchronized. This flat process ensures all necessary steps are taken without architectural ambiguity.

Best Practices

• Explicit over Implicit: Always set the stack explicitly when starting work on a new component.
• Verification: Run get_active_stack after setting to confirm the change was persisted.
• Memory Sync: Ensure scripts successfully update the Memory MCP graph for cross-session consistency.