Dop, Awk-like processing for JSON/YAML/TOML with Lua
jq and yq already do this—Lua embedding is nice but doesn't justify switching from established tools.

Zero-code MCP exposure for existing CLI commands, but MCP adoption is still niche.
Backend developers, DevOps engineers, and anyone running local services who want to expose CLI commands to AI assistants.
MCP SDK (Python/TypeScript/Go) · Anthropic's MCP specification · Cursor IDE integrations
jq and yq already do this—Lua embedding is nice but doesn't justify switching from established tools.
Direct QEMU control without libvirt overhead brings Docker Compose DX to VMs.
MCP servers as shell pipes—clever abstraction, but only matters if you're already in the MCP ecosystem.
YAML runbook executor with destructive-step approval gates, exposable as MCP tools directly.
Loop driver + 15 slash commands for Claude Code, but orchestration over integration.
Yet another one-command AI stack when Ollama and Jan already exist.