skills/core/claude-md-routing/SKILL.md
Keep CLAUDE.md lean (≤200 lines) by routing verbose reference content to separate files instead of inlining it.
npx skillsauth add bereniketech/claude_kit claude-md-routingInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
CLAUDE.md is loaded into every session as the system prompt. Everything in it costs tokens every single turn. Bloated CLAUDE.md = context rot faster, worse performance, higher cost.
Rule: Keep CLAUDE.md at 150–200 lines maximum. If it grows past that, trim or route.
| Inline (in CLAUDE.md) | Route to a separate file |
|----------------------|--------------------------|
| Project goal (2–3 lines) | Style guide (colors, fonts, component rules) |
| Tech stack (1 line) | Business context (user personas, domain glossary) |
| Key folder paths | API reference docs |
| Active skill imports (@ paths) | Historical decisions / ADRs |
| Critical rules Claude must never forget | Onboarding / setup steps |
| Pointers to routed files | Long checklists |
.claude/
├── CLAUDE.md ← lean — 150 lines max
├── context/
│ ├── style-guide.md ← design tokens, component conventions
│ ├── business.md ← domain, personas, terminology
│ ├── api-reference.md ← endpoint list and shapes
│ └── decisions.md ← architectural decisions log
└── project-config.md ← skill list, stack, agent roster
Point to routed files with a single line so Claude knows where to look without loading the content:
## Reference Files
- Style guide: `.claude/context/style-guide.md`
- Business context: `.claude/context/business.md`
- API reference: `.claude/context/api-reference.md`
Claude reads these on demand. They don't load into every session — only when the task requires them.
Run this when CLAUDE.md exceeds 150 lines:
@ import lines are not routable — they must stay in CLAUDE.md for Claude Code to resolve them. Keep skill imports short and targeted: 2–6 skills per project, only the ones active for current work.
Rule: Don't import a skill just because it exists. Import it when the project needs it, remove it when the work is done.
testing
AUTHORIZED USE ONLY: This skill contains dual-use security techniques. Before proceeding with any bypass or analysis: > 1.
testing
Provide comprehensive techniques for attacking Microsoft Active Directory environments. Covers reconnaissance, credential harvesting, Kerberos attacks, lateral movement, privilege escalation, and domain dominance for red team operations and penetration testing.
development
Detects missing zeroization of sensitive data in source code and identifies zeroization removed by compiler optimizations, with assembly-level analysis, and control-flow verification. Use for auditing C/C++/Rust code handling secrets, keys, passwords, or other sensitive data.
development
Comprehensive guide to auditing web content against WCAG 2.2 guidelines with actionable remediation strategies.