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

# seri

> A cross-platform coding-agent CLI that ships as one binary and asks before it writes.

seri is a coding agent you run in your terminal. It ships as a single `seri` binary with no
runtime to install, and runs natively on Windows, macOS and Linux, with no WSL2 and no Docker.

```sh theme={null}
seri
```

That is the whole invocation. seri opens a full-screen session, and everything you do with it
happens inside that session as a slash command.

## What it is built for

seri is **code-first, not code-only.** Coding is what it does today, and it is what this release
is built for. The tools it ships are file, search and shell tools. The loop, the session store
and the permission model are deliberately not bounded to a repository, so evaluate seri today as
a coding agent.

## What defines it

**It asks before it writes.** A permission gate sits under every tool call, on every OS.
`approve-each` is the default, so the first write of a session stops and waits for you. Approving
a tool "always" is scoped to that one tool, in that one project.

**Its edits are reversible.** seri records your worktree into a shadow git repository as it
works, independent of your own branch, so undoing its edits never touches your commit history.
Shell commands are covered when they look destructive.

**Six providers, three ways to pay.** Groq, OpenRouter, Anthropic, OpenAI, Google and xAI, with
your own key, a Grok subscription, or a hosted seri account. Switching mid-session keeps the
conversation.

**It extends without a source change.** Agents, skills, rules, MCP servers and hooks are all
files you drop in a directory. seri picks them up at the next start.

**It checks its own work.** Point seri at your project's own check command and it runs after
every successful `write_file`, handing the diagnostics back to the model in the same turn.

## Start here

<CardGroup cols={2}>
  <Card title="Install seri" icon="download" href="/get-started/installation">
    One command per platform, and what each installer touches.
  </Card>

  <Card title="Your first session" icon="terminal" href="/get-started/quickstart">
    Guided setup, a first task, an approval, and undoing it.
  </Card>

  <Card title="Providers and models" icon="route" href="/get-started/models">
    Six providers, three credential classes, and rerouting.
  </Card>

  <Card title="Slash commands" icon="list" href="/reference/commands">
    Every command in one table.
  </Card>
</CardGroup>

## Extend it

<CardGroup cols={2}>
  <Card title="Subagents" icon="sitemap" href="/guides/subagents">
    Named roles with their own context, and agent files you write.
  </Card>

  <Card title="Skills" icon="scroll" href="/guides/skills">
    Procedures that run in the session's own context.
  </Card>

  <Card title="Rules" icon="book" href="/guides/rules">
    Standing instructions, always on or scoped by glob.
  </Card>

  <Card title="MCP servers" icon="plug" href="/guides/mcp">
    Remote tools, with OAuth and a trust step.
  </Card>
</CardGroup>
