Sbh

Name: Sbh
Author: boshu2

boshu2/agentops

23 installs
416 repo stars
Updated August 5, 2026
boshu2/agentops

sbh (Storage Ballast Helper) is a Claude Code skill and CLI daemon that prevents disk-full failures on AI coding hosts using ballast files, artifact scanning, and pressure monitoring.

About

SBH (Storage Ballast Helper) is a disk-pressure defense tool for AI coding workloads. It prevents disk-full failures through pre-allocated ballast files that release instantly, an artifact scanner that scores stale files for cleanup, and a daemon that monitors pressure across mounts including /tmp, /dev/shm, and swap. A developer uses it via CLI or systemd/launchd to check free space, run emergency cleanup, and see pressure sources in a TUI dashboard.

Releases pre-allocated ballast files in milliseconds to free disk space
Scans and scores stale build artifacts for cleanup
Runs a daemon with predictive disk-pressure monitoring and a TUI dashboard

Sbh by the numbers

23 all-time installs (skills.sh)
Ranked #807 of 1,039 Cloud & Infrastructure skills by installs in the Skillselion catalog
Data as of Aug 5, 2026 (Skillselion catalog sync)

At a glance

sbh capabilities & compatibility

Capabilities: devops
Use cases: devops
Platforms: Linux · macOS

From the docs

What sbh says it does

Disk-pressure defense for AI coding workloads.

SKILL.md

Prevents disk-full disasters via ballast files, artifact scanning, and predictive pressure monitoring.

SKILL.md

Pre-allocated sacrificial files — released in milliseconds, no scanning needed.

SKILL.md

npx skills add https://github.com/boshu2/agentops --skill sbh

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

[![Listed on Skillselion](https://skillselion.com/badge/skills/boshu2/agentops/sbh.svg)](https://skillselion.com/skills/boshu2/agentops/sbh)

Installs	23
repo stars	★ 416
Last updated	August 5, 2026
Repository	boshu2/agentops ↗

What it does

Prevent disk-full failures on AI coding hosts via ballast files, artifact scanning, and pressure monitoring.

Who is it for?

Guarding disk space on machines running AI coding agents and CI builds

Skip if: Placing ballast on /tmp instead of the same mount as the pressure source

When should I use this skill?

Disk is full or low on space, or you need cleanup, ballast, or pressure monitoring

What you get

Instant space recovery via ballast release plus scored artifact cleanup before the disk hits full.

By the numbers

Default 10 x 1 GiB = 10 GiB ballast pool
TUI dashboard with 7 screens
5-factor cleanup scoring adding to 1.0

Files

SKILL.mdMarkdownGitHub ↗

SBH — Storage Ballast Helper

Prevents disk-full disasters via ballast files, artifact scanning, and predictive pressure monitoring. Three-pronged: ballast (instant space), scanner (stale artifacts), special locations (/tmp, /dev/shm, swap).

Quick Check

sbh status                     # Pressure level + free space
sbh status --json | jq '.pressure'  # Machine-parseable
sbh check --need 5G            # "Do I have 5 GB free?"
sbh check --predict 30         # "Will I run out in 30 min?"

Exit codes: 0 = healthy, 1 = pressure, 2 = error.

---

Daemon

sbh daemon                          # Foreground (debugging)
systemctl --user start sbh          # Systemd user scope
sbh install --systemd --user --auto # Install + start (Linux)
sbh install --launchd --auto        # Install + start (macOS)
sbh install --wizard                # Guided interactive setup

Signals: SIGHUP = reload config, SIGUSR1 = force scan now, SIGTERM = graceful stop.

---

Ballast

Pre-allocated sacrificial files — released in milliseconds, no scanning needed.

sbh ballast status             # Per-volume inventory
sbh ballast provision          # Create/rebuild pool
sbh ballast release 3          # Free 3 files NOW
sbh ballast replenish          # Rebuild after pressure passes

Defaults: 10 x 1 GiB = 10 GiB. Ensure ballast dir is on same mount as pressure source.

---

Scanning & Cleanup

sbh scan /data/projects --top 20       # Rank artifacts by score
sbh clean /data/projects --dry-run     # Preview what would go
sbh clean --target-free 50G --yes      # Delete until 50 GB free

Scoring: Location (.25) + Name (.25) + Age (.20) + Size (.15) + Structure (.15) = 1.0.

---

Protection

sbh protect /path              # .sbh-protect marker (subtree)
sbh unprotect /path            # Remove marker

Config globs: scanner.protected_paths. Hard vetoes (always enforced): .git/ dirs, open files, age < 10 min, non-writable parents.

---

Emergency Recovery

Zero-write mode for near-100% full disks. No config file needed.

sbh emergency /data --yes              # Aggressive cleanup NOW
sbh emergency --target-free 10G        # Stop at 10 GB recovered

---

Observability

sbh dashboard                  # TUI: 7 screens (1-7 to jump)
sbh stats --window 24h         # Activity over last 24 hours
sbh blame --top 10             # Top 10 pressure sources
sbh explain --id <ID>          # Why was this decision made?

---

Configuration

Config: ~/.config/sbh/config.toml | Env: SBH_ prefix | Fallback: /etc/sbh/config.toml

sbh config show                # Current values
sbh config validate            # Check constraints
sbh config set KEY VALUE       # Change a value
sbh tune --apply --yes         # Auto-tune for this system

---

Anti-Patterns

Don't	Do Instead
Ballast on `/tmp`	`paths.ballast_dir` on same mount as pressure source
Daemon as root, CLI as user	`--user` scope — avoids state file permission mismatch
Skip pre-build check	`sbh check --need 10G` in CI/hook
Delete `.sbh-protect` by hand	`sbh unprotect /path`
Wait for Red to act	Act at Yellow — agent swarms escalate fast
`min_file_age_minutes = 0`	Keep >= 5 to protect in-flight writes