Claude Code

Set up

curl -fsSL hanzo.sh | sh
hanzo login
hanzo --claude

Give it a task as one quoted argument to run it headless: hanzo --claude "fix the failing test in api/users". claude-code and cc are accepted spellings of claude.

On your Claude plan (the default)

Claude Code uses its own login in ~/.claude, and its model calls go to Anthropic on your Claude plan. The wrapper adds:

• The Hanzo toolset. The native hanzo-mcp that hanzo.sh installs beside hanzo is attached with --mcp-config for the run, through the CLI's own proxy, so Claude Code starts without waiting on it. Claude Code can then hand work to Enso (llm), Kai (kai_decide) and Hanzo's search tools, billed to your Hanzo plan. See Hand work to Enso and Kai.
• The session is registered in the cloud with its machine and directory. Pass --no-link to keep it local.
• The summary at the end names every toolset call and the model it reached.

Fully on Hanzo (--cloud)

hanzo --claude --cloud
hanzo --claude --cloud --model enso-ultra

• Model calls go to https://api.hanzo.ai. The wrapper points Claude Code at it with ANTHROPIC_BASE_URL and sends your Hanzo token as the bearer in ANTHROPIC_AUTH_TOKEN, and clears any ANTHROPIC_API_KEY in your shell so one credential is in play. The token reaches the process through its environment, never its arguments.
• The model is enso-auto unless you set another with --model, ANTHROPIC_MODEL or model in ~/.hanzo/settings.json. --model zen6 runs Zen 6, the open-weights model. Claude Code is handed a claude-* name so it grants the model's real context window, and the wrapper maps that name back to the Hanzo model before the request leaves the process.
• Claude Code's own slots are Enso tiers: Haiku, which runs background work and routine turns, is enso-flash; Sonnet is enso-pro; Opus and Fable are enso-ultra. Enso has what each tier costs and which models answer it.
• Usage is metered into your org at api.hanzo.ai, so it appears in the cloud whichever machine you ran on. Your org's cloud tools are added beside the toolset, because the process already holds your token.
• Claude Code's home for the run is ~/.hanzo/claude, so its saved model choice and your Hanzo settings do not overwrite each other.

What does not load

• A repository's own .mcp.json and .claude/settings*.json hooks do not load, because a repository is untrusted and its hooks would run with your credentials in the environment. --project-mcp opts in for a repository you trust.
• With --ask the tool asks before each action. By default the wrapper starts Claude Code with --dangerously-skip-permissions, as autoApprove in ~/.hanzo/settings.json says.

See usage

hanzo console          # press 6 for Savings
hanzo usage summary --range 7d

How to read the summary →