What's New
New skill: reviewing-cc-config — Review your Claude Code configuration for context efficiency, signal density, and anti-patterns.
Derived from Anthropic's "Effective Context Engineering for AI Agents" and the official Claude Code Best Practices, this skill applies 16 review rules across 4 categories to audit your entire Claude Code setup:
Review Rubric (16 rules)
| Category | Rules | What It Checks |
|---|---|---|
| Context Budget (CB-*) | CB-STARTUP, CB-FORK, CB-PROGRESSIVE, CB-COMPACT-SAFE | Session startup token budget, context isolation, just-in-time loading |
| Signal Density (SD-*) | SD-CLAUDE-MD, SD-DESCRIPTION, SD-TOOL-MINIMAL, SD-RETURN | Non-derivable instructions, routing precision, minimal tools, condensed output |
| Architecture (AR-*) | AR-HOOK-DETERMINISTIC, AR-SKILL-DEMAND, AR-MODEL-MATCH, AR-VERIFY, AR-ISOLATION | Hooks vs CLAUDE.md placement, on-demand skills, model routing, verification |
| Anti-Patterns (AP-*) | AP-TRIGGER-OVERLAP, AP-SCOPE-UNBOUNDED, AP-OVER-SPECIFIED | Trigger ambiguity, unbounded agents, config bloat |
How It Works
- Discovery — scans both standard (
.claude/skills/) and plugin (plugins/*/) layouts - Token Budget — estimates session startup cost and flags bloat (>4K tokens = HEAVY)
- Parallel Review — spawns up to 4 specialized agents reviewing CLAUDE.md, skills, agents, and hooks
- Report — structured findings by severity with cross-cutting analysis and top-5 improvements
Usage
/reviewing-cc-config # Review all config components
/reviewing-cc-config skills # Review only skills
/reviewing-cc-config claude-md # Review only CLAUDE.md files
/reviewing-cc-config hooks agents # Review hooks and agents
/reviewing-cc-config all --fix # Review everything, then apply fixes
Stats
- Skill count: 31 → 32 (dev-tools: 14 → 15)
- All 32 skills exported for Codex CLI, Gemini CLI, and AGENTS.md
- Co-located
RUBRIC.mdwith full rule definitions and source citations
Full Changelog: v1.5.0...v1.6.0