NUILab Agentic AI
Docs / Tools & Frameworks / Claude Code
Tools & Frameworks
Claude Code
Anthropic's command-line coding agent — the frontier lane of the stack, run under amux for parallel, isolated sessions.
Claude Code is a coding agent that runs in the terminal: you give it a task in a repository and it reads, edits, runs commands, and iterates. In this stack it is the frontier lane — the model you reach for on the hard 20% of the work (see local and frontier models ). amux runs it in named, isolated sessions, so you can have several going at once.
Install and sign in
Install it once and log in, so it can launch unattended under amux:
curl -fsSL https://claude.ai/install.sh | bash
claude # first run: sign in
The native installer keeps itself up to date. After that, amux drives the same claude binary —
that is the supported way to automate a subscription: you are using your own account through its own
client, not a backend pretending to be an API (see subscriptions, APIs, and local models
).
Flags worth knowing
You pass these through amux-run
,
after the session name:
| Flag | Effect |
|---|---|
--model opus / --model sonnet | Pick the model. |
--continue | Resume the last conversation in this directory. |
--resume | Choose which past conversation to resume. |
--append-system-prompt "…" | Add standing instructions for the session. |
For the full reference, see the Claude Code docs (opens in new tab) .
Next
- Run it under amux: Run agents on one machine .
- How its access works: subscriptions, APIs, and local models .
- Steer it with standing rules: give an agent standing instructions .
- Extend it: skills, MCP, and extending an agent .
- Make the bottom line useful: customize the status line .
- The cheaper lane for routine work: local and frontier models .
Source: content/tools/claude-code.md · maintained in the nuilab-agenticai repository.