> ## Documentation Index
> Fetch the complete documentation index at: https://cli-docs.relai.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# All commands

> The complete public relai surface. Read-only commands are safe to run freely; others write files, upload metadata, remove records, or change code.

## Machine & setup

| Command             | Use it to                                              |
| ------------------- | ------------------------------------------------------ |
| `relai info`        | Print installed CLI information. *(read-only)*         |
| `relai setup`       | Configure machine-level preferences & credentials.     |
| `relai update`      | Update or reinstall the CLI.                           |
| `relai plugins`     | Install local coding-agent plugin marketplaces.        |
| `relai completions` | Generate shell completion scripts. *(read-only)*       |
| `relai report`      | Send a log file & feedback to RELAI for investigation. |

See [setup](/cli/setup) or [report](/cli/report) for flags and examples.

## The learning loop

The core cycle — see [The learning loop](/learning-loop) for how these stages fit together.

| Command          | Use it to                                            |                            |
| ---------------- | ---------------------------------------------------- | -------------------------- |
| `relai init`     | Register a project & generate simulator support.     | [details →](/cli/init)     |
| `relai simulate` | Run selected learning environments.                  | [details →](/cli/simulate) |
| `relai optimize` | Improve the agent against environments & benchmarks. | [details →](/cli/optimize) |

## Authoring

| Command              | Use it to                                  |                                  |
| -------------------- | ------------------------------------------ | -------------------------------- |
| `relai learning-env` | Create & upload learning environments.     | [details →](/cli/learning-env)   |
| `relai evaluator`    | Create, upload & remove global evaluators. | [details →](/cli/evaluator)      |
| `relai benchmark`    | Register & list benchmark suites.          | [details →](/cli/benchmark)      |
| `relai tag`          | Group learning environments with tags.     | [details →](/cli/tag-and-memory) |
| `relai memory`       | Read & update Markdown-backed memory.      | [details →](/cli/tag-and-memory) |

## Project inputs

Use a root-level `.relaiignore` file to keep selected project files out of RELAI backend generation inputs and optimizer worktrees. See [Controlling project inputs with `.relaiignore`](/cli/init#controlling-project-inputs-with-relaiignore) for scope, examples, and explicit-path failure behavior.

## Getting help

Use `--help` at any level for command-specific flags, and `--version` for the installed version.

```sh theme={"system"}
relai --help
relai simulate --help
relai learning-env create --help
relai --version
```

## Diagnostic logs

Each `relai` invocation writes a bounded, structured diagnostic log to
`~/.relai/logs/`. Logs record command lifecycle, safe high-level actions,
validation and repair attempts, backend outcomes, warnings, and failures. CLI
arguments are represented by flag names only, paths are project-relative, and
subprocesses retain only operation, attempt, status, outcome, and duration.
Logs intentionally exclude credentials, option values, authorization headers,
request bodies, generated source, and all subprocess output.

Each invocation is limited to 500 events or 256 KiB. Logs older than 30 days
are removed, and the directory is pruned to 100 MiB. When asking for support,
share only the relevant log after reviewing it for project names or other
information you do not want to disclose.