What problem does it solve? Keeping the same personal coding rules synchronized across Claude Code, Codex, OpenCode, Pi, TRAE CN, and Grok Build is tedious and error-prone, since each agent stores user-level instructions in different files and formats. This Skill detects which agents are installed, resolves their effective instruction targets, and writes your exact role text safely without overwriting unrelated configuration. ## Core Features & Use Cases - Multi-agent detection: Probes executables, environment variables (CLAUDE_CONFIG_DIR, CODEX_HOME, OPENCODE_CONFIG_DIR, PI_CODING_AGENT_DIR), and config roots to determine which supported agents are actually installed and where their active instruction files live. - Safe dual-mode writes: Writes dedicated rule files for Claude/Grok/TRAE CN, or inserts a managed rules-install marker block into shared aggregate files like Codex AGENTS.md, preserving all unrelated content. - Backup and validation: Creates timestamped backups before every change, re-reads targets to verify exact content, and reports installed, unchanged, skipped, and unresolved agents with restart instructions. - Use Case: You switch between Codex and Claude Code daily and want the same coding style rules active in both. Provide your role text once, confirm the detected install scope, and the Skill updates both agents with backups and verification. ## Quick Start Use the rules-install skill to detect my installed coding agents, show me the effective target paths, and install my global role text into the agents I confirm.
