agent-skills-setuplisted

Migrate AI assistant context between IDEs — MCP servers, rules/instructions, skills, slash commands, agents, hooks, and memory. Resolves only the user-specified source/target paths, converts supported formats, previews every change, preserves or backs up existing data according to an explicit strategy, and verifies the result. WHEN TO USE: ONLY when the user explicitly asks to migrate, move, transfer, copy, convert, or sync AI assistant context between IDEs, for example “migrate MCP config”, “move skills from Cursor to Claude”, or “transfer rules from Windsurf to Cursor”. Never write without a reviewed --dry-run and explicit approval. DO NOT ACTIVATE ON: incidental mentions of MCP, skills, rules, or IDE names; format questions; debugging requests; “how do I…” / “what is…” questions; or any prompt that does not explicitly request a cross-IDE migration. When source, target, objects, or scope are unclear, ask before reading configuration.
Luckycat133/skills-repo · ★ 2 · AI & Automation · score 72

Install: claude install-skill Luckycat133/skills-repo

# AI IDE Context Migration Migrate only the context the user names. Treat path, scope, schema, secrets, and conflicts as part of the migration contract rather than assuming two IDEs store equivalent files. ## Load references progressively Keep the first response fast and focused: 1. Read only the source and target sections in [references/ide-registry.md](references/ide-registry.md). 2. Use [references/ide-paths.json](references/ide-paths.json) or **scripts/smart-ide-migration.sh --print-path** for deterministic lookup. 3. Read other registry sections only when a referenced compatibility surface is actually involved. Do not load the complete registry by default. The registry is the detailed source for platform paths, schemas, precedence, verification commands, and manual boundaries. The script is the executable contract. If prose and executable behavior disagree, stop and report the drift. ## Security model - Resolve only the two IDEs and optional workspace the user specifies. Never enumerate the home directory or auto-discover every installed IDE. - Default objects are low risk: **skills,rules,prompts**. MCP, config, project, agents, hooks, and memory require explicit selection. - Start with **--dry-run**. It may parse the selected source but creates no workspace or target output and never prints raw configuration values. - A write requires explicit approval and **--yes**. A non-interactive run without it exits before any target write. - Literal creden