What problem does it solve? When an AI agent needs to query a codebase through GitNexus MCP, the repository must first have a fresh, valid index. This Skill removes the operational guesswork of creating indexes for new repositories, updating stale ones after code changes, and repairing corrupted indexes (broken full-text search, vector dimension mismatches, incomplete query results). ## Core Features & Use Cases - Environment Detection and Runner Resolution: Checks for a project-local runner, global CLI, or npx fallback before running any indexing command, and verifies the target is a Git repository. - Status-Driven Indexing Decisions: Uses gitnexus status and gitnexus list to decide between first-time creation, incremental update, forced full rebuild, or targeted repair, avoiding wasteful full re-indexing. - Repair and Validation Workflow: Maps specific failures to fixes (--force, --repair-fts, --embeddings --drop-embeddings) and verifies index freshness after every operation. - Use Case: You hand an AI agent a newly cloned monorepo. The agent detects no existing index, configures .gitnexusignore to exclude node_modules and dist, runs gitnexus analyze, and confirms via gitnexus status that the index is fresh before answering code questions. ## Quick Start Ask the AI to create or update the GitNexus code index for the current repository and confirm the index is fresh.