Skip to main content

​
General Questions

​
How do I check which agents/commands are installed?

/plugin list everything-claude-code@everything-claude-code
This shows all available agents, commands, and skills from the plugin.

​
Can I use only some components (e.g., just agents)?

Yes. Use manual installation and copy only what you need: 
# Just agents
cp agents/*.md ~/.claude/agents/

# Just rules
cp -r rules/common/* ~/.claude/rules/
Each component is fully independent.

​
Does this work with Cursor / OpenCode / Codex?

Yes. ECC is cross-platform:
  • Cursor: Pre-translated configs in .cursor/. See README Cursor IDE Support section.
  • OpenCode: Full plugin support in .opencode/. See README OpenCode Support section.
  • Codex: First-class support with adapter drift guards and SessionStart fallback.
  • Claude Code: Native — this is the primary target.

​
How do I contribute a new skill or agent?

See CONTRIBUTING.md. Short version:
  1. Fork the repo
  2. Create your skill in skills/your-skill-name/SKILL.md
  3. Or create an agent in agents/your-agent.md
  4. Submit a PR with clear description

​
Installation & Setup

​
What’s the difference between plugin install and manual install?

Plugin install:
  • One command: /plugin install everything-claude-code@everything-claude-code
  • Auto-updates with plugin system
  • Namespaced commands: /everything-claude-code:plan
  • Rules must still be installed manually (Claude Code limitation)
Manual install:
  • Copy each component manually
  • Full control over what’s installed
  • Shorter command names: /plan
  • Requires manual updates

​
Why do I need to install rules manually?

Claude Code plugin system does not support distributing rules automatically (upstream limitation). You must copy them manually: 
cp -r rules/common/* ~/.claude/rules/
cp -r rules/typescript/* ~/.claude/rules/
Or use the installer: 
./install.sh typescript

​
What’s the minimum Claude Code version required?

v2.1.0 or later Check your version: 
claude --version
Earlier versions have incompatible hook handling.

​
Hooks

​
My hooks aren’t working / I see “Duplicate hooks file” errors

Do NOT add a "hooks" field to .claude-plugin/plugin.json. Claude Code v2.1+ automatically loads hooks/hooks.json from installed plugins. Explicitly declaring it causes duplicate detection errors. See:

​
How do I disable a specific hook?

Override in ~/.claude/settings.json: 
{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Write",
        "hooks": [],
        "description": "Override: allow all .md file creation"
      }
    ]
  }
}

​
Can I add my own hooks?

Yes. See Hooks README for examples and recipes.

​
Context & Tokens

​
My context window is shrinking / Claude is running out of context

Cause: Too many MCP servers eating your context. Each MCP tool description consumes tokens from your 200k window, potentially reducing it to ~70k. Fix: Disable unused MCPs per project in .claude/settings.json: 
{
  "disabledMcpServers": ["supabase", "railway", "vercel"]
}
Guidelines:
  • Keep under 10 MCPs enabled
  • Keep under 80 tools active

​
How do I reduce token costs?

Apply token optimization settings in ~/.claude/settings.json: 
{
  "model": "sonnet",
  "env": {
    "MAX_THINKING_TOKENS": "10000",
    "CLAUDE_AUTOCOMPACT_PCT_OVERRIDE": "50",
    "CLAUDE_CODE_SUBAGENT_MODEL": "haiku"
  }
}
Impact:
  • ~60% cost reduction (opus → sonnet)
  • ~70% reduction in hidden thinking costs
  • Better quality in long sessions (compact at 50% vs 95%)
See Token Optimization.

​
When should I use /compact vs /clear?

/clear:
  • Between unrelated tasks
  • Free and instant
  • Completely resets context
/compact:
  • At logical task breakpoints:
    • After research, before implementation
    • After milestone, before next phase
    • After debugging, before feature work
  • Costs tokens but preserves context
  • Don’t use mid-implementation

​
Skills & Agents

​
How do I know which skills are active?

Claude Code loads all skills in ~/.claude/skills/ and .claude/skills/. List them: 
ls -la ~/.claude/skills/
ls -la .claude/skills/

​
Which skills should I install?

For new users: Start with core bundle:
  • .agents/skills/* — Engineering, evals, verification
  • skills/search-first/ — Research-first workflow
For specific stacks:
  • TypeScript/React: frontend-patterns, backend-patterns, coding-standards
  • Python/Django: django-patterns, django-tdd, python-patterns
  • Go: golang-patterns, golang-testing
  • Java/Spring Boot: springboot-patterns, springboot-tdd
For workflows:
  • TDD: tdd-workflow, language-specific testing skills
  • Security: security-review, use /security-scan
  • Database: postgres-patterns, clickhouse-io, jpa-patterns

​
How are agents different from skills?

Agents:
  • Specialized subagents for delegation
  • Defined in agents/*.md
  • Invoked via commands or auto-delegation
  • Have specific tool restrictions
  • Examples: code-reviewer, tdd-guide, planner
Skills:
  • Knowledge modules and workflow definitions
  • Defined in skills/*/SKILL.md
  • Loaded as context, not invoked directly
  • Provide reference material
  • Examples: coding-standards, frontend-patterns

​
Package Managers

​
How do I change my package manager?

/setup-pm
Or set environment variable: 
export CLAUDE_PACKAGE_MANAGER=pnpm
Or use script: 
node scripts/setup-package-manager.js --global pnpm
See Package Managers Reference.

​
What package managers are supported?

  • npm (default fallback)
  • pnpm
  • yarn (classic and berry)
  • bun

​
Why is the wrong package manager being used?

Check detection order: 
node scripts/setup-package-manager.js --detect
Priority:
  1. Environment variable (CLAUDE_PACKAGE_MANAGER)
  2. Project config (.claude/package-manager.json)
  3. package.json packageManager field
  4. Lock file detection
  5. Global config (~/.claude/package-manager.json)
  6. First available package manager
Override with highest priority (environment variable).

​
Security

​
How do I scan for security issues?

/security-scan
# Or:
npx ecc-agentshield scan
See AgentShield Reference.

​
What does AgentShield check?

5 categories:
  1. Secrets detection (14 patterns)
  2. Permission auditing
  3. Hook injection analysis
  4. MCP server risk profiling
  5. Agent config review
Scans: CLAUDE.md, settings.json, .claude.json, hooks/, agents/, skills/

​
How do I fix hardcoded secrets?

Wrong: 
# CLAUDE.md
API_KEY=sk-1234567890abcdef
Right: 
# CLAUDE.md
API_KEY=${OPENAI_API_KEY}
Then set in shell: 
export OPENAI_API_KEY=sk-1234567890abcdef
Or use secret manager (AWS Secrets Manager, 1Password, etc.).

​
Advanced

​
Can I use this with Agent Teams?

Yes, but be aware: Agent Teams spawns multiple context windows. Each teammate consumes tokens independently. Only use for tasks where parallelism provides clear value:
  • Multi-module work
  • Parallel reviews
  • Independent research tasks
For simple sequential tasks, subagents are more token-efficient.

​
How do I use continuous learning?

Two versions: v1 (continuous-learning): 
/learn   # Extract patterns mid-session
v2 (continuous-learning-v2): 
/instinct-status        # View learned instincts
/instinct-import <file> # Import instincts
/instinct-export        # Export instincts
/evolve                 # Cluster instincts into skills
See Continuous Learning Guide.

​
How do I create project-specific skills?

Use the template: 
cp -r skills/project-guidelines-example ~/.claude/skills/my-project/
Edit ~/.claude/skills/my-project/SKILL.md with project-specific:
  • Architecture decisions
  • Naming conventions
  • Deployment procedures
  • Team workflows

​
What’s the search-first skill?

Enforces research-before-coding workflow:
  1. Search existing code first
  2. Check documentation
  3. Only then implement
Prevents:
  • Reinventing existing solutions
  • Inconsistent patterns
  • Missing context
Use with Exa MCP for web research.

​
Troubleshooting

​
Where can I find more troubleshooting help?

See Troubleshooting Reference for:
  • Plugin issues
  • Hook problems
  • Context/token issues
  • Package manager errors
  • Installation problems

​
How do I report a bug?

  1. Search GitHub Issues
  2. If not found, open new issue with:
    • ECC version
    • Claude Code version
    • OS and Node.js version
    • Steps to reproduce
    • Error messages/logs

​
How do I request a feature?

Open a GitHub Discussion in the Ideas category. Include:
  • Use case (what you’re trying to achieve)
  • Proposed solution (how it might work)
  • Alternatives considered