What problem does it solve? AI-assisted coding sessions often suffer from scope drift, repeated rework, and unclear failures, but teams lack a structured way to understand why. This Skill performs forensic root cause analysis on Antigravity session artifacts to explain what changed, who or what caused it, and what should improve next time. ## Core Features & Use Cases - Session Forensics: Reads conversation artifacts (task.md, implementation_plan.md, walkthrough.md, resolved snapshots) from the Antigravity brain directory to reconstruct what happened in each session. - Root Cause Classification: Assigns primary causes such as SPEC_AMBIGUITY, HUMAN_SCOPE_CHANGE, REPO_FRAGILITY, AGENT_ARCHITECTURAL_ERROR, VERIFICATION_CHURN, or LEGITIMATE_TASK_COMPLEXITY, each with evidence and confidence levels. - Severity Scoring & Hotspots: Scores sessions 0-100 across completion, replanning, scope instability, and rework dimensions, and clusters friction by file or subsystem. - Use Case: After a week of AI-assisted development with heavy rework, run this Skill to generate a session_analysis_report.md showing that most churn came from ambiguous opening prompts, then rewrite your prompt templates accordingly. ## Quick Start Analyze my recent Antigravity sessions and generate a root cause report explaining why sessions required rework.
