Add ncc utility skill: host operational and health CLI

[zivisaiah]

Type of Change

• Feature skill - adds a channel or integration (source code changes + SKILL.md)
• Utility skill - adds a standalone tool (code files in .claude/skills/<name>/, no source changes)
• Operational/container skill - adds a workflow or agent skill (SKILL.md only, no source changes)
• Fix - bug fix or security fix to source code
• Simplification - reduces or simplifies source code
• Documentation - docs, README, or CONTRIBUTING changes only

Description

Adds ncc (NanoClaw Control), a host-side operational and health CLI, as a sibling to ncl. Where ncl does central-DB entity CRUD (agent groups, wirings, roles), ncc covers the operational surface ncl does not: host service lifecycle, a component health scan, and visibility into scheduled work.

Why. There is no shipped tool for host service control or health inspection. Service management is documented only as raw launchctl/systemctl commands, and /debug is instruction-only with no executable scanner. ncc fills that gap.

How it works. A standalone Python 3 script (standard library only). It reads the session and central SQLite DBs directly and shells out to launchctl/systemctl/docker, so it keeps working even when the host service is down, which is when health and service start matter most. Cross-platform: launchd on macOS, systemd (--user) on Linux, both auto-detected (the unit is discovered, not hardcoded). It mirrors ncl's UX: subcommands, --json, table output, exit codes.

Commands:

• ncc service status|start|stop|restart
• ncc health [--json] scans the service, Docker daemon, central DB integrity, host disk, the OneCLI gateway, each agent container (autocompact thrashing, zombie processes, heartbeat freshness), recent permanent delivery failures, and overdue scheduled tasks. Exits 2 on a hard FAIL, 0 otherwise (WARN is advisory).
• ncc tasks [--json] running containers plus pending ad-hoc tasks
• ncc crons [--json] recurring scheduled tasks
• ncc --version, ncc --help, ncc <command> --help

How it was tested. Installed via the SKILL.md steps on macOS (launchd) against a live install and exercised every subcommand, confirmed the --json output parses, and verified the health exit codes (0 on OK/WARN, 2 on FAIL) and that the symlinked binary resolves the checkout from any working directory. The Linux/systemd path was validated in a Python 3.12 Linux container: platform detection, unit discovery, status parsing, correct systemctl invocation, and the SQLite/disk/heartbeat reads. Real-daemon validation on a live systemd host is still pending. On Linux the service is a systemctl --user unit, so ncc must be run as the service's owning user; this is documented in the skill's troubleshooting.

Usage. ncc health, ncc service restart, ncc crons --json. Install steps are in the SKILL.md.

For Skills

• SKILL.md contains instructions, not inline code (code goes in separate files)
• SKILL.md is under 500 lines
• I tested this skill on a fresh clone