> ## Documentation Index
> Fetch the complete documentation index at: https://zuno-fb55ec99.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> The interactive shell workflow.

```
npm install -g @zunocli/cli
export OPENAI_API_KEY=sk-...
export ZUNO_ARBITRUM_RPC_URL=https://...
zuno

create my zuno wallet
show my zuno wallet
fund my zuno wallet
inspect my positions
inspect position <tokenId>
create a position with 0.05 ETH passively
recommend what I should do with this position
show me the diff
simulate it
approve it
apply it
refresh pools
show alerts
```

At startup, Zuno prints a short env notice when optional keys are missing. Add the keys to your shell profile (`~/.zshrc`, `~/.bashrc`, etc.) if you want them available every time.

| intent    | summary                                                         |
| --------- | --------------------------------------------------------------- |
| wallet    | Create, inspect, fund, and check the Turnkey-backed Zuno wallet |
| positions | List Uniswap v4 positions held by the Zuno wallet               |
| inspect   | Print one position snapshot: range, current tick, status        |
| create    | Open a brand-new position via the four-agent debate             |
| recommend | Rebalance an existing position via the four-agent debate        |
| diff      | Re-render before/after range and inventory changes              |
| simulate  | Preview deterministic execution steps before signing            |
| approve   | Human approval before any state-changing action                 |
| apply     | Policy check, Turnkey signing, and transaction submission       |
| monitor   | Inspect monitor setup and read alerts from the worker           |

Plans are persisted to `~/.zuno/plans/<planId>.json` so `zuno diff` works
across sessions.

## Notes

* The CLI is itself an AXL peer for recommendation flows. Direct reads,
  diffs, simulation, and wallet state stay outside AXL.
* If the mesh is not online, recommendation runs the same four-agent
  debate in-process. If no LLM is available either, the deterministic
  `recommendPlan` math runs as a final fallback. Plans persist either way.
* Continuous monitoring runs as `pnpm monitor`; the shell stays the control
  surface and reads alerts with `show alerts`.
* Output is plain ANSI. No colour-stripping needed for piping.