Migrating from Claude Code

Me Write Code can reuse Claude Code authoring formats while storing its own config under ~/.mewrite/agent/ and project config under .mewrite/.

TL;DR

# 1. Install
npm install -g @zhachory1/mewrite-code

# 2. Copy config
mkdir -p ~/.mewrite/agent
cp -r ~/.claude/commands ~/.mewrite/agent/
cp -r ~/.claude/skills ~/.mewrite/agent/
cp -r ~/.claude/agents ~/.mewrite/agent/
cp ~/.claude/settings.json ~/.mewrite/agent/settings.json    # hooks + statusLine

# 3. Project-scope
cp -r .claude .mewrite   # optional project-scope import

# 4. Keep CLAUDE.md or add AGENTS.md; Me Write Code reads both.

# 5. MCP — already standard
#    .mcp.json works as-is.

# 6. Run
mewrite

What maps directly

Claude Code	Me Write Code	Notes
~/.claude/settings.json	~/.mewrite/agent/settings.json	Hooks + statusLine compatible schema
~/.claude/commands/*.md	~/.mewrite/agent/commands/*.md	Frontmatter is a superset
~/.claude/skills/<name>/SKILL.md	~/.mewrite/agent/skills/<name>/SKILL.md	Identical
~/.claude/agents/<name>.md	~/.mewrite/agent/agents/<name>.md	Frontmatter is a superset
.mcp.json	.mcp.json	Same path; no change
CLAUDE.md	CLAUDE.md or AGENTS.md	Me Write Code reads both, layered
Auto-Memory	cavemem	Different backend; same UX

Differences worth knowing

Memory

Claude Code uses Auto-Memory with ~/.claude/projects/<slug>/memory/MEMORY.md. Me Write Code uses cavemem by default and falls back to FilesProvider when Cavemem is unavailable. To bridge:

mewrite memory sync --from claude

This imports MEMORY.md and per-fact files as cavemem observations. Going forward, if you keep both Claude Code and Me Write Code running in the same project, mewrite-code reads the first 200 lines of MEMORY.md on every session start.

Models

Claude Code is Anthropic-only. Me Write Code is provider-agnostic. After migrating, you can:

mewrite --model openai/gpt-5-codex
mewrite --model claude-sonnet-4   # default behavior matches Claude Code

Cost

By default Caveman Mode compression is on, which Claude Code doesn't have. Expect tool-output token consumption to drop ~85%. If something looks off, bisect with:

/cave off

Permissions and hooks

Me Write Code supports plan mode, approval mode, checkpoints, and beta native sandboxing. PreToolUse hooks can deny, ask, or allow tool calls via Claude Code-compatible hook output.

Confirming the migration worked

mewrite doctor                    # general health
# inside the TUI:
/hooks list                       # all hooks loaded
mewrite skills list               # all skills loaded
mewrite agents list               # all subagents loaded
mewrite mcp doctor                # MCP servers reachable

If any of these report mismatches, open an issue — we treat Claude Code parity as a CI gate.

Why not just use Claude Code?

• Token efficiency. Caveman Mode, tool-output budgets, read deduplication, and prompt-cache-friendly sessions reduce context waste.
• Provider flexibility. Use ChatGPT Plus, Copilot, Gemini, or any OpenAI-compatible endpoint.
• Session branching. /tree, /fork — no major competitor has this.
• MIT. No vendor lock-in; self-host the daemon.

If none of those matter to you, stay on Claude Code — it's a fine product.