2026年3月11日 · 7 分読み取り
Unused skills, agents, commands, hooks, and rules are not passive clutter. They add search noise, distort AI context, and make cross-repo reuse riskier unless teams clean them deliberately.
- vibesmith
- ai-coding
- component-operations
記事を読む
2026年3月11日 · 4 分読み取り
Cursor, Claude Code, MCP, and GitLab Duo can accelerate one repo quickly. The real systems problem begins when standards, dependencies, and context have to survive across many repos.
- multi-repo
- ai-coding
- vibesmith
記事を読む
2026年3月10日 · 4 分読み取り
Cursor Rules and AGENTS.md solve different parts of the same problem. Teams that use both deliberately get more durable AI coding context across repositories.
記事を読む
2026年3月9日 · 4 分読み取り
Claude Code is increasingly powerful, but multi-repo teams get the most leverage from shared hooks, project commands, and durable review gates rather than one-off prompts.
記事を読む
2026年3月8日 · 3 分読み取り
MCP matters because it turns tool integrations into a shared protocol. It does not solve standards, naming, approvals, or dependency visibility by itself.
- mcp
- ai-infrastructure
- workflows
記事を読む