From pm-multimodels
Configure and synchronize repositories so Codex and Cursor reuse canonical Claude Code guidance, global and project skills, and commands. Use when setting up a target repository, translating Claude-specific tools, diagnosing configuration drift, resolving generated-file conflicts, or installing the global watcher.
How this skill is triggered — by the user, by Claude, or both
Slash command
/pm-multimodels:configure-multimodelThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use the deterministic engine instead of manually copying or rewriting configuration.
Use the deterministic engine instead of manually copying or rewriting configuration.
Identify the repository root and confirm CLAUDE.md is the canonical project file.
Run:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/pm-multimodels" configure <repo>
Review the dry-run report with the user.
Stop on every unmanaged destination or modified generated artifact.
Ask the user to choose symlink or copy for repository artifacts.
Ask separately before changing .gitignore or replacing an existing AGENTS.md.
Apply only after approval by adding --apply.
Use sync --global to translate ~/.claude/skills and ~/.claude/commands into the managed cache, then expose them to Codex and Cursor with symlinks.
Never symlink raw Claude skills directly into another platform. Generate a platform adapter first, then symlink the adapter directory.
Use watch for a foreground watcher. On macOS, use watch --install-launch-agent only after explicit approval.
Read references/translation-contract.md when reviewing adapters or adding mappings.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub mbjornson/pm-multimodel --plugin pm-multimodels