Appendices

Appendix A: File Locations Reference

Quick reference for where Claude Code stores files and configuration.

Windows

Component	Location
npm global bin	`C:\Users\<username>\AppData\Roaming\npm`
Node.js install	`C:\Program Files\nodejs`
Claude data directory	`C:\Users\<username>\.claude\`
Claude config file	`C:\Users\<username>\.claude.json`
Log files	`%APPDATA%\Claude\logs\`
MCP config	`C:\Users\<username>\.claude.json` (`mcpServers` field)
Session data	`C:\Users\<username>\.claude\local\`
Downloads/cache	`C:\Users\<username>\.claude\downloads\`

Quick Access (PowerShell):

# Open Claude data directory
explorer "$env:USERPROFILE\.claude"

# Open config file
notepad "$env:USERPROFILE\.claude.json"

# View logs
Get-Content "$env:APPDATA\Claude\logs\mcp*.log" -Wait -Tail 50

macOS

Component	Location
npm global bin	`/usr/local/bin` or `$(npm config get prefix)/bin`
Node.js install	`/usr/local/bin/node` (Homebrew) or `/opt/homebrew/bin/node` (M1/M2)
Claude data directory	`~/.claude/`
Claude config file	`~/.claude.json`
Log files	`~/Library/Logs/Claude/`
MCP config	`~/.claude.json` (`mcpServers` field)
Session data	`~/.claude/local/`
Downloads/cache	`~/.claude/downloads/`

Quick Access:

# Open Claude data directory
open ~/.claude

# Edit config file
code ~/.claude.json  # VS Code
# or
nano ~/.claude.json  # Terminal editor

# View logs
tail -f ~/Library/Logs/Claude/mcp*.log

Linux

Component	Location
npm global bin	`/usr/local/bin` or `~/.npm-global/bin`
Node.js install	`/usr/bin/node`
Claude data directory	`~/.claude/`
Claude config file	`~/.claude.json`
Log files	`~/.local/share/claude/logs/` or `~/.cache/claude/logs/`
MCP config	`~/.claude.json` (`mcpServers` field)
Session data	`~/.claude/local/`
Downloads/cache	`~/.claude/downloads/`

Quick Access:

# Open Claude data directory
cd ~/.claude

# Edit config file
nano ~/.claude.json
# or
vim ~/.claude.json

# View logs
tail -f ~/.local/share/claude/logs/mcp*.log

Project-Level Files

These are the same across all platforms:

File/Directory	Location	Purpose	Commit to Git?
`CLAUDE.md`	Project root	Project memory (team)	✅ Yes
`.claude/CLAUDE.md`	Project root	Personal memory	❌ No
`.claude/settings.json`	Project root	Hook configuration	✅ Yes
`.claude/settings.local.json`	Project root	Personal permissions	❌ No
`.claude/agents/`	Project root	Custom agents	✅ Yes (team)
`.claude/commands/`	Project root	Custom commands	✅ Yes (team)
`.claude/hooks/`	Project root	Event hooks	✅ Yes (team)
`.claude/skills/`	Project root	Knowledge modules	✅ Yes (team)
`.claude/rules/`	Project root	Auto-load rules	✅ Yes (team)
`.claude/.serena/`	Project root	Serena MCP index	❌ No

Environment Variables

Set these in your shell profile (~/.zshrc, ~/.bashrc, or Windows System Properties):

Variable	Purpose	Example
`ANTHROPIC_API_KEY`	API authentication	`sk-ant-api03-...`
`ANTHROPIC_BASE_URL`	Alternative API endpoint	`https://api.deepseek.com/anthropic`
`ANTHROPIC_MODEL`	Default model	`claude-sonnet-4-20250514`
`ANTHROPIC_SMALL_FAST_MODEL`	Fast model for simple tasks	`claude-haiku-4-20250514`
`BASH_DEFAULT_TIMEOUT_MS`	Bash command timeout	`60000`
`ANTHROPIC_AUTH_TOKEN`	Alternative auth token	Your auth token
`CLAUDE_CODE_DISABLE_1M_CONTEXT`	Disable 1M context window support (v2.1.50+)	`true`
`CLAUDE_CODE_SIMPLE`	Fully minimal mode: disables skills, agents, MCP, hooks, CLAUDE.md loading (v2.1.50+)	`true`

Finding Your Paths

Can’t find npm global bin?

# Universal command
npm config get prefix

# Should output something like:
# macOS/Linux: /usr/local or ~/.npm-global
# Windows: C:\Users\<username>\AppData\Roaming\npm

Can’t find Claude executable?

# macOS/Linux
which claude

# Windows (PowerShell)
where.exe claude

# Windows (CMD)
where claude

Can’t find log files?

# Run Claude with debug and check output
claude --debug 2>&1 | grep -i "log"

Recommended .gitignore

Add these to your project’s .gitignore:

# Claude Code - Personal/Local
.claude/settings.local.json
.claude/CLAUDE.md
.claude/.serena/
.claude/local/

# Claude Code - Team (DO commit these)
# .claude/agents/
# .claude/commands/
# .claude/hooks/
# .claude/skills/
# .claude/settings.json

# API Keys
.env
.env.local
.env.*.local
*.key

# OS Files
.DS_Store
Thumbs.db

Appendix B: FAQ

Claude Code vs ClawdBot: What’s the Difference?

Question: Both tools use “Claude” in their name and I’ve seen buzz about both recently. Are they competitors? Which should I choose?

Short answer: They serve completely different use cases. Not competitors—complementary tools for different audiences.

Detailed comparison:

Aspect	Claude Code	ClawdBot
Interface	Terminal/CLI + IDE integration (VS Code, Cursor, etc.)	Messaging apps (WhatsApp, Telegram, Discord, Signal, iMessage)
Primary audience	Software developers, DevOps, tech leads	Everyone (personal assistants, smart home, knowledge workers)
Core use case	Software development (code generation, refactoring, debugging, architecture)	Personal automation, task management, smart home control, 24/7 assistance
Access model	Local terminal session, requires being at computer or SSH	Remote access via messaging apps from any device (phone, watch, tablet)
Installation	`npm install -g claude-code` (simple CLI install)	Self-hosted (Docker/VPS ~$5/month + LLM API costs)
Architecture	CLI tool + MCP servers + hooks/skills/agents system	Agent + Gateway + Skills + Memory (4-component architecture)
Smart home	Not covered (development-focused)	✅ Native Home Assistant integration, monitoring, automation
Code development	✅ Core use case (pair programming, code review, refactoring)	Possible but not the primary focus
Pricing	Anthropic API usage (pay-as-you-go, ~$3-15/project typical)	Open-source (free) + infrastructure (VPS ~$5/mo + LLM API costs)
Ideal for	Writing code, reviewing PRs, debugging, architectural decisions	Daily reminders, email management, calendar, monitoring, personal tasks

When to choose Claude Code:

You’re a developer working in terminal/IDE
You need pair programming, code reviews, refactoring help
Your workflow is codebase analysis and software architecture
You want deep IDE integration and git workflow automation

When to choose ClawdBot:

You want a 24/7 personal assistant accessible from your phone
You need smart home automation (Home Assistant, IoT devices)
Your use cases are: task management, reminders, email/calendar, monitoring
You want messaging app interface (WhatsApp, Telegram, etc.)

Can you use both?:

Yes! They complement each other well:

ClawdBot: Handles daily personal automation, monitoring, reminders
Claude Code: Handles software development work when at your computer

Example workflow:

ClawdBot (on phone): "Remind me to review the PR when I'm at my desk"
  → Get notification
Claude Code (at desk): "Review the authentication PR, check for security issues"
  → Deep code analysis with full codebase context

Quick decision tree:

What's your primary goal?
├─ Write/review code → Claude Code
├─ Personal automation/smart home → ClawdBot
└─ Both? → Use both (they don't conflict)

Resources:

ClawdBot: https://clawd.bot/ | GitHub | Setup guide | Community adoption analysis
Claude Code: This guide | GitHub

Community confusion:

Common misconceptions we’ve seen:

❌ “ClawdBot is Claude Code but with messaging interface” → False. Different architectures, different use cases.
❌ “I need to choose one or the other” → False. They complement each other.
❌ “ClawdBot is a fork of Claude Code” → False. Independent projects with different creators.

Final note: This comparison reflects Jan 2026 state of both tools. ClawdBot has documented strong community adoption (5,600+ social mentions, use cases ranging from smart home to radio decoding). Both are evolving rapidly. Check official documentation for latest capabilities.

Can Product Managers use Claude Code?

Short answer: Yes, but consider your primary workflow first.

Code-adjacent PMs (reviewing technical specs, PRDs, architecture feasibility):

✅ Claude Code CLI is appropriate for technical validation workflows
Example: Granola meeting notes → ChatPRD generation → Claude Code refinement
Use case: Auditing technical feasibility, generating specs from PRDs

Non-coding PMs (strategy, research, stakeholder management):

⚠️ Claude Code CLI overhead not justified
✅ Better fit: Claude Desktop (see Cowork Guide)
Use case: Research synthesis, stakeholder communication, roadmap planning

Tool Stack Example (via Stilyan Mitrev, Head of Product StableLab):

Meeting capture: Granola + Wispr Flow (dictation)
PRD generation: ChatPRD → Claude Code review
UI prototyping: v0 → Claude Code feasibility check
Workflow pattern: Base context project + specialized projects per domain

Reality check: PM workflows with Claude Code are an emerging area with limited community validation. We currently have 1 practitioner report (the source practitioner noted they tried Claude Code but didn’t adopt it long-term). If you’re a PM using Claude Code successfully, contribute your workflow to help the community.

See also:

AI Ecosystem Guide — Complementary tools (Granola, Wispr Flow, ChatPRD, v0)
Cowork Guide — Claude Desktop for non-technical PMs
Design-to-Code Workflow — PM perspective on Figma MCP

Can I continue a session from a different project folder?

Short answer: Not with native `—resume`, but manual filesystem operations work reliably.

The limitation: Claude Code’s `—resume` command is scoped to the current working directory by design. Sessions are stored at `~/.claude/projects//` where the path is derived from your project’s absolute location. Moving a project or forking a session to a new folder breaks the resume capability.

Why this design?: Sessions store absolute file paths, project-specific context (MCP server configurations, `.claudeignore` rules, environment variables). Cross-folder resume would require path rewriting and context validation, which isn’t implemented yet.

Workaround - Manual migration (recommended):

```bash

When moving a project folder

cd ~/.claude/projects/ mv — -old-location-myapp- -new-location-myapp-

When forking sessions to a new project

cp -n ./-source-project-/*.jsonl ./-target-project-/ cp -r ./-source-project-/subagents ./-target-project-/ 2>/dev/null || true

cd /path/to/target/project && claude —continue ```

⚠️ Migration risks:

Hardcoded secrets/credentials may not transfer correctly
Absolute paths in session context may break
MCP server configurations may differ between projects
`.claudeignore` rules are project-specific

Community automation: The claude-migrate-session skill by Jim Weller automates this process, but has limited testing (0 stars/forks as of Feb 2026). Manual approach is safer.

Detailed guide: See Session Resume Limitations & Cross-Folder Migration for complete workflow and edge cases.

Related: GitHub issue #1516 tracks community requests for native cross-folder support.

Appendix C: Resource Evaluation Process

This guide systematically evaluates external resources (tools, methodologies, articles, frameworks) before integration to maintain quality and prevent noise.

Evaluation Methodology

5-Point Scoring System (Critical → Low):

Score	Action	Timeline
5	Critical - Integrate immediately	<24h
4	High Value - Integrate within 1 week	1 week
3	Moderate - Integrate when time available	Flexible
2	Marginal - Minimal mention or skip	-
1	Low - Reject	-

Full methodology: See docs/resource-evaluations/README.md

Current Evaluations

14 documented assessments in docs/resource-evaluations/:

Methodologies: GSD (Get Shit Done), Vibe Coding patterns
Tools: Worktrunk, AST-grep, SE-CoVe plugin
Content: Boris Cherny Cowork video, ClawdBot Twitter analysis
Research: Prompt repetition papers, self-improvement skills
And more: Nick Jensen plugins, Wooldridge productivity stack

Browse all evaluations: docs/resource-evaluations/

Why Public Evaluations?

Transparency: Contributors can see exactly why resources were:

✅ Integrated (score 3+): Added to guide with attribution
⚠️ Mentioned (score 2): Brief reference without deep coverage
❌ Rejected (score 1): Documented reason for exclusion

Quality Control: Technical review + challenge phase by specialized agents ensures objectivity and prevents marketing hype from influencing decisions.

Community Contribution: Evaluation template available in docs/resource-evaluations/README.md for suggesting new resources with systematic assessment.

Appendix D: Myths vs Reality

This section addresses common misconceptions about Claude Code circulating in online communities, social media, and discussions.

❌ Myth: “Claude Code has hidden features you can unlock with secret flags”

Reality: All public features are documented in the official CHANGELOG.

What people confuse:

Progressive rollout ≠ Hidden features: Anthropic uses feature flags for staged deployment (standard industry practice)
Experimental features ≠ Secrets: Features like TeammateTool exist but are clearly marked as experimental/unstable
Community discovery ≠ Hacking: When users discover unreleased features in compiled code, that’s exploration, not “unlocking secrets”

The truth about feature flags:

Flag	Purpose	Status
`CLAUDE_CODE_ENABLE_TASKS=false`	Revert to old TodoWrite system (v2.1.19+)	Official migration path
TeammateTool flags	Progressive deployment of multi-agent orchestration	Experimental, unstable
Other internal flags	Quality assurance, A/B testing, staged rollout	Not meant for end users

Best practice: Read the CHANGELOG and official release notes. Features become public when they’re stable and documented. Using experimental features via workarounds can cause:

Data loss or corruption
Crashes and instability
Incompatibility with future versions
Loss of official support

Red flags to watch for (signs of misinformation):

“Hidden feature that will blow your mind!”
“Secret trick the devs don’t want you to know”
No citation of official sources (CHANGELOG, docs, GitHub issues)
FOMO language: “If you’re not using this, you’re falling behind”
Dramatic claims: “This changes everything” without evidence

❌ Myth: “Tasks API allows fully autonomous parallel agents”

Reality: The Tasks API (v2.1.16+) enables coordination of parallel work, but agents are not autonomous.

What Tasks API actually does:

Creates a shared task list with dependency tracking
Allows main session + sub-agents to coordinate work
Persists tasks across sessions for resumption
Notifies sessions when tasks complete

What it does NOT do:

❌ Automatically spawn agents for each task
❌ Create self-organizing “swarms” of independent agents
❌ Enable agents to make decisions without human approval
❌ Replace your need to manage and direct the work

How parallel execution actually works:

You → Create tasks with TaskCreate
You → Spawn sub-agents with Task tool (explicit action)
You → Sub-agents work independently in parallel
You → Sub-agents return summaries
You → Coordinate next steps

Sources:

CHANGELOG v2.1.16 - Official task management release
Section 2.6 - Task Management - Full documentation

❌ Myth: “Claude Code is 100x faster than other AI coding tools”

Reality: Performance depends on task complexity, model choice, and how you use the tool. No tool is universally “100x faster.”

What affects speed:

Model selection: Haiku (fast) vs Sonnet (balanced) vs Opus (thorough)
Context management: Effective use of sub-agents, MCP servers, strategic compaction
Prompt quality: Clear requirements vs vague instructions
Task complexity: Simple refactoring vs architectural analysis

Honest comparison (typical use cases):

Task	Claude Code	Other Tools	Winner
Simple edits (typos, formatting)	~5-10s	~5-10s	≈ Tie
Multi-file refactoring	30-60s	60-120s	Claude Code (2x)
Complex architecture analysis	2-5min	5-15min	Claude Code (3x)
Learning curve (first week)	Moderate	Varies	Depends on tool

The truth: Claude Code is powerful and efficient, but claims of “100x faster” are marketing hyperbole. Real advantage comes from:

Deep context window (200K tokens)
Smart sub-agent system (prevents context pollution)
MCP ecosystem (specialized tools)
Strong system prompts (high-quality outputs)

✅ Reality: What Makes Claude Code Actually Special

Documented, verifiable strengths:

Context Window: 200K tokens (~150K words) - one of the largest in the industry
Sub-Agent System: Isolated context windows prevent pollution during exploration
MCP Ecosystem: 100+ community servers for specialized tasks
Permission System: Granular control over tool access and dangerous operations
CLI-First Design: Terminal integration, git workflows, IDE compatibility
Transparent Pricing: Pay-as-you-go, no subscriptions, predictable costs
Active Development: Weekly releases with community-driven features

Sources: All claims verifiable in official documentation and CHANGELOG.

How to Spot Reliable Information

✅ Trust these sources:

Official Claude Code documentation
GitHub CHANGELOG
GitHub Issues with Anthropic staff responses
Community resources citing official sources (like Claudelog.com)
This guide (with 14 evaluated resources and clear sourcing)

❌ Be skeptical of:

Social media posts with no sources
“Secret tricks” without CHANGELOG references
Percentage claims without benchmarks (“50% faster”, “10x productivity”)
Dramatic language designed to create FOMO
Content that discourages reading official docs

Contributing to This Section

Found a new myth circulating online? Open an issue with:

The myth/misconception
Where you saw it (platform, approximate reach)
Why it’s misleading (with sources)

We’ll evaluate and add it to this section if it meets quality criteria.

About This Guide

End of Guide

Author: Florian BRUNIAUX | Founding Engineer @Méthode Aristote

Written with: Claude (Anthropic) - This guide was collaboratively written with Claude Code, demonstrating the tool’s capabilities for technical documentation.

Inspired by:

Claudelog.com - An excellent resource for Claude Code tips, patterns, and advanced techniques that served as a major reference for this guide.
ykdojo/claude-code-tips - Practical productivity techniques that informed keyboard shortcuts, context handoffs, and terminal workflow optimizations in sections 1.3, 2.2, and 10.2.
Nick Tune’s Coding Agent Development Workflows - Advanced workflow patterns integrated in sections 3.1, 7.1, 9.3, and 9.10.

License: CC BY-SA 4.0 - Feel free to use, adapt, and share with attribution.

Contributions: Issues and PRs welcome.

Last updated: January 2026 | Version: 3.32.2