Verify the Claude Agent Pack installation and current project scaffolding in one pass. Runs check-readiness.sh and check-updates.sh, reports all failures and outdated items together, and offers to run remediation. Use before starting work on a project or when something seems off after a git pull. Trigger this when someone says: check my setup, is everything installed, something is not working, verify the pack, is my installation up to date, check if Claude Agent Pack is ready, run a health ch...
Installs into .claude/skills of the current project.
Are you the author of System Check?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/chuckplayer-system-check)
---
name: system-check
description: "Verify the Claude Agent Pack installation and current project scaffolding in one pass. Runs check-readiness.sh and check-updates.sh, reports all failures and outdated items together, and offers to run remediation. Use before starting work on a project or when something seems off after a git pull. Trigger this when someone says: check my setup, is everything installed, something is not working, verify the pack, is my installation up to date, check if Claude Agent Pack is ready, run a health check. Do NOT use to initialize a new project — use /setup-project instead."
---
# System Check
Verify the Claude Agent Pack installation and current project scaffolding in a single pass.
## 1. Locate the pack
Find the pack directory by locating `scripts/check-readiness.sh`. The pack is typically cloned to a fixed location (e.g., `~/claude-agent-pack`). If you cannot locate it, ask the user where they cloned the pack.
## 2. Run both checks
Run readiness and update checks, passing the current project directory to readiness:
```bash
bash <pack-dir>/scripts/check-readiness.sh <project-dir>
bash <pack-dir>/scripts/check-updates.sh
```
Capture the full output of both.
## 3. Interpret readiness results
For each `[!!]` failure in check-readiness output:
| Failure | Remediation |
|---------|-------------|
| `~/.claude directory` | Claude Code is not installed — direct the user to install it |
| Agents not installed | `bash <pack-dir>/install.sh` |
| Skills not installed | `bash <pack-dir>/install.sh` |
| `CLAUDE.md` missing | `bash <pack-dir>/scripts/setup-project.sh <project-dir>` |
| `docs/CONVENTIONS.md` missing | `bash <pack-dir>/scripts/setup-project.sh <project-dir>` |
| `docs/MEMORY-WRITING.md` missing | `bash <pack-dir>/scripts/setup-project.sh <project-dir>` |
| `memory/<subdir>/` missing | `bash <pack-dir>/scripts/setup-project.sh <project-dir>` |
Group multiple project-scaffolding failures under a single `setup-project.sh` invocation.
## 4. Interpret update results
The update check reports three states for each agent and skill:
| State | Meaning |
|-------|---------|
| `[ok]` | Installed and matches the pack source |
| `[!!] (outdated)` | Installed but differs from the pack source |
| `[--] (not installed)` | Not installed at all |
List any outdated or missing agents and skills. If anything is outdated or missing, the fix is:
```bash
bash <pack-dir>/install.sh
```
## 5. Report and offer to remediate
Present a unified summary:
- All readiness failures with their remediation commands
- All outdated or missing agents/skills
- If everything is clean, confirm the pack and project are fully set up and ready
Offer to run the remediation commands on the user's behalf if they confirm.
## Gotchas
- **Pack directory not found:** If `scripts/check-readiness.sh` cannot be located, ask the user where they cloned the agent pack before doing anything else. Do not guess at common locations — the path varies by machine.
- **check-readiness.sh fails with a permission error:** The script must be executable. If it fails with `Permission denied`, ask the user to run `chmod +x <pack-dir>/scripts/*.sh` and retry.
- **Multiple project-scaffolding failures:** If `CLAUDE.md`, `docs/CONVENTIONS.md`, `docs/MEMORY-WRITING.md`, and `memory/` subdirectories are all missing, a single `setup-project.sh` call fixes all of them. Group these under one remediation command rather than listing them separately.
- **Outdated agents/skills vs. missing agents/skills:** Both are fixed by `install.sh`, but the distinction matters for the user's awareness. Report them separately so the user understands whether this is a fresh install or an update.
- **User declines remediation:** If the user does not want to run remediation, report the failures clearly and stop. Do not attempt workarounds or partial fixes.