Claude Code Setup

Custom configurations and skills for Claude Code.

Contents

• Status Line - Enhanced status bar with context tracking
• Skills - Reusable prompt templates for common workflows

---

Status Line

A custom status line that shows:

• Model name (with accent color)
• Current directory
• Git branch with uncommitted file count
• Sync status with upstream (ahead/behind/synced + last fetch time)
• Visual context bar showing token usage percentage
• Error count from project's error database (optional)
• Last user message on second line

Screenshot

Claude Opus 4.5 | 📁myproject | 🔀main (2 files uncommitted, synced 5m ago) | ▄░░░░░░░░░ 12% of 200k tokens
💬 can you fix the login bug

Installation

1. Copy context-bar.sh to ~/.claude/:

cp context-bar.sh ~/.claude/context-bar.sh
chmod +x ~/.claude/context-bar.sh

2. Add the status line configuration to your ~/.claude/settings.json:

{
  "statusLine": {
    "command": "/path/to/your/home/.claude/context-bar.sh",
    "type": "command"
  }
}

Replace /path/to/your/home with your actual home directory path (e.g., /Users/yourname on macOS or /home/yourname on Linux).

Color Themes

The script supports 10 color themes. Change the COLOR variable at the top of the script:

• gray (monochrome)
• orange
• blue (default)
• teal
• green
• lavender
• rose
• gold
• slate
• cyan

Requirements

• jq - for parsing JSON
• git - for repository status
• Bash 4+

---

Skills

Skills are reusable prompt templates that can be invoked in Claude Code. Place them in ~/.claude/commands/ or ~/.claude/skills/.

Reinforce (/reinforce)

Self-reinforcing code quality workflow with dual-group adversarial review.

Requires: Codex MCP server configured in your settings.json:

"mcpServers": {
  "codex": {
    "command": "npx",
    "args": ["@anthropic/codex-mcp"]
  }
}

Launches 6 parallel agents (3 Opus + 3 Codex) to review your code from different angles:

• Bugs & Logic - Edge cases, race conditions, null handling
• Security - Injection, XSS, auth bypass, input validation
• Performance/UX/QA - N+1 queries, loading states, accessibility

The workflow:

1. Both groups review the same code independently
2. Findings are displayed side-by-side
3. Consensus issues (found by both) are prioritized
4. All issues are fixed in severity order

# Review recent changes
/reinforce

# Review specific files
/reinforce src/api/auth.ts src/components/Login.tsx

Thinking Critically (/thinking-critically)

Senior staff-level architectural thinking framework.

A 15-step analysis process for planning systems, evaluating changes, or analyzing problems:

1. Clarify the Problem
2. Explore the Design Space (3-5 distinct approaches)
3. Tradeoff Analysis
4. Failure-First Thinking (top 10 failure modes)
5. Boundaries & Invariants
6. Observability & Control
7. Reversibility & Entropy Control
8. Adversarial Review
9. AI Delegation Assessment
10. Decision Summary
11. Dependency Mapping
12. Migration & Rollout Strategy
13. Security Threat Modeling (STRIDE)
14. Blast Radius Analysis
15. Operational Runbook Preview

# Analyze a system design
/thinking-critically "Add real-time notifications to the dashboard"

# Evaluate a proposed change
/thinking-critically "Migrate from PostgreSQL to MongoDB"

Installing Skills

Copy the skill files to your Claude Code commands directory:

cp skills/reinforce.md ~/.claude/commands/reinforce.md
cp skills/thinking-critically.md ~/.claude/commands/thinking-critically.md

Or for the skills directory structure:

mkdir -p ~/.claude/skills/thinking-critically
cp skills/thinking-critically.md ~/.claude/skills/thinking-critically/SKILL.md

---

License

MIT