Claude Code maintenance, diagnostics, and usage analytics. Check inotify limits, clean caches, diagnose startup failures, manage skills directories, and track token usage, costs, and rate limit patterns across projects.
Scanned 9/11/2026
Install to Claude Code
npx -y skills add grahama1970/agent-skills --skill ops-claude --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Ops Claude?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/grahama1970-ops-claude)More formats (shields.io, HTML) on the badges page.
---
name: ops-claude
description: >
Claude Code maintenance, diagnostics, and usage analytics. Check inotify limits,
clean caches, diagnose startup failures, manage skills directories, and track
token usage, costs, and rate limit patterns across projects.
allowed-tools: Bash, Read
triggers:
- claude won't start
- claude crashes
- claude out of memory
- inotify limit
- file watcher error
- ENOSPC
- clean claude cache
- claude diagnostics
- ops claude
- claude usage
- how much claude am I using
- claude costs
- claude savings
- claude roi
- token usage
- claude report
- claude spending
- rate limit correlation
metadata:
short-description: Claude Code maintenance, diagnostics, and usage analytics
provides:
- ops-claude
composes:
- debug-pdf
- scillm
- create-figure
- analytics
- rate-limit-recovery
- memory
- task-monitor
- agentic-evals
disciplines:
- observability-operations
- developer-tooling
---
> STOP. READ THIS ENTIRE SKILL.MD BEFORE CALLING ANY ENDPOINT.
# Ops-Claude Skill
Maintenance, diagnostics, and usage analytics for Claude Code installations.
## Supported IDEs/CLIs
| IDE | Skills Path | User-level |
|-----|-------------|------------|
| Claude Code | `.claude/skills` | `~/.claude/skills` |
| Pi Agent | `.pi/skills` | `~/.pi/skills` |
| Codex | `.codex/skills` | `~/.codex/skills` |
| KiloCode | `.kilocode/skills` | `~/.kilocode/skills` |
| GitHub Copilot | `.github/skills` | - |
| Antigravity/Gemini | `.gemini/skills` | `~/.gemini/skills` |
| Agent (generic) | `.agent/skills` | `~/.agent/skills` |
## Common Issues Addressed
| Error | Cause | Fix Command |
|-------|-------|-------------|
| `ENOSPC: System limit for number of file watchers` | inotify exhaustion | `./run.sh fix-inotify` |
| `JavaScript heap out of memory` | Too many files in skills | `./run.sh clean-skills` |
| `Allocation failed` | Node.js heap limit | `./run.sh fix-heap` |
| Claude hangs on startup | Corrupted cache | `./run.sh clean-cache` |
## Maintenance Commands
### diagnose
Run full diagnostics and report issues.
```bash
./run.sh diagnose # Check everything
./run.sh diagnose --fix # Check and fix automatically
```
### status
Show current resource usage and limits.
```bash
./run.sh status
```
### fix-inotify / fix-heap
Fix inotify limits or Node.js heap size.
```bash
./run.sh fix-inotify # Requires sudo
./run.sh fix-heap
```
### clean-skills / clean-cache / clean-all
Clean .venv/__pycache__ from skills, or Claude Code caches.
```bash
./run.sh clean-skills
./run.sh clean-cache
./run.sh clean-all
```
### gitignore
Add/update .gitignore in all skills directories.
```bash
./run.sh gitignore
```
## Usage Analytics Commands
Requires `ccusage` CLI (already installed). Calculates API-equivalent costs for
Max plan usage — shows what you'd pay without the subscription.
### report
Usage dashboard with token volumes and API-equivalent costs.
```bash
./run.sh report --daily --days 7 # Daily view (default)
./run.sh report --monthly # Monthly totals
./run.sh report --daily --days 7 --json # JSON for /create-figure
```
### savings
Max plan ROI calculator.
```bash
./run.sh savings --monthly # Current month breakdown
./run.sh savings --json # JSON output
```
### correlate
Rate limit pattern analysis — cross-references usage with rate limit windows.
```bash
./run.sh correlate --days 14
```
### insights
Trend analysis, anomaly detection, and project breakdowns.
```bash
./run.sh insights --days 14
```
### top-projects
Per-project usage breakdown.
```bash
./run.sh top-projects --days 7 --limit 10
```
## JSON Output & Figure Integration
All analytics commands support `--json` for automation. JSON output includes
`figure_data` compatible with `/create-figure`:
```bash
# Generate line chart of daily costs
./run.sh report --daily --days 7 --json > /tmp/claude_usage.json
.pi/skills/create-figure/run.sh line /tmp/claude_usage.json --key figure_data.line
```
## Nightly Report
A scheduled job (`claude-usage-nightly`) runs at 6:00am daily:
- Generates yesterday's usage summary
- Compares against 7-day rolling average
- Detects anomalies (>2x average)
- Stores summaries to `~/.pi/memory/ops-claude/`
## Proactive Usage
Run `./run.sh diagnose` when:
- Claude Code fails to start
- Claude Code crashes with memory errors
- After syncing skills across projects
- System feels sluggish with many projects open
Run `./run.sh report` to:
- Track daily/monthly token consumption
- Justify Max plan ROI
- Identify high-usage projects
- Correlate rate limits with usage patterns
## Common Mistakes
### WRONG: Manually cleaning Claude Code caches without diagnostics
```bash
rm -rf ~/.claude/cache/ # may delete needed state
```
### RIGHT: Use the skill's clean commands
```bash
./run.sh clean-cache # safe cleanup of Claude Code caches
./run.sh clean-skills # removes .venv/__pycache__ from skills
```
### WRONG: Ignoring inotify limit errors (ENOSPC)
```
ENOSPC: System limit for number of file watchers reached
# Claude Code hangs or crashes
```
### RIGHT: Fix inotify limits immediately
```bash
./run.sh fix-inotify # requires sudo, increases watcher limit
```
### WRONG: Not using --json for automation integration
```bash
./run.sh report --daily --days 7 # human-readable, not parseable
```
### RIGHT: Use --json for /create-figure integration
```bash
./run.sh report --daily --days 7 --json > /tmp/usage.json
.pi/skills/create-figure/run.sh line /tmp/usage.json --key figure_data.line
```
## References (retrieve on demand — do not vendor)
External docs drift; cite the canonical URLs and fetch them when needed
with `/context7` (library docs) or `/fetcher` (any URL/PDF) rather than
caching stale copies. Verified reachable (HTTP 200) 2026-08-24.
- Claude Code documentation: <https://docs.claude.com/en/docs/claude-code/overview>
- llms.txt (LLM-friendly doc index): <https://docs.claude.com/llms.txt>
- llms-full.txt (expanded LLM index): <https://docs.claude.com/llms-full.txt>
```bash
skills/context7/run.sh "claude code cli hooks settings"
skills/fetcher/run.sh "https://docs.claude.com/en/docs/claude-code/overview"
```
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!