# Clarvis

> A terminal workspace for running, supervising, and extending coding agents with explicit safety, context, and human review.

The links below are clean Markdown alternatives to the public documentation. Follow only the pages needed for the current task.

## Start here

- [Installation](https://clarvis.dev/installation.md): Install, verify, update, or remove a portable Clarvis beta without requiring a development toolchain.
- [Getting started](https://clarvis.dev/getting-started.md): Start Clarvis in a project, complete the first-run setup, choose a safety posture, run a useful task, and learn the few controls that keep you in charge of the session.
- [Daily use](https://clarvis.dev/guide/daily-use.md): Run focused tasks, steer work in progress, inspect results, and keep sessions organized without losing control of the workspace.
- [Providers and models](https://clarvis.dev/guide/providers-and-models.md): Connect an API or subscription, choose which models are available, and set the model and reasoning effort Clarvis uses by default.
- [Safety and control](https://clarvis.dev/guide/safety.md): Choose where commands may run, who reviews risk, and which workspace configuration Clarvis is allowed to activate.

## Plan and operate

- [Plans](https://clarvis.dev/guide/plans.md): Decide whether work needs a plan review, keep completed plans as an audit trail, or discard them automatically after successful work.
- [Worktrees](https://clarvis.dev/guide/worktrees.md): Give Clarvis a dedicated Git branch and checkout without leaving your primary working tree.
- [Agents](https://clarvis.dev/guide/agents.md): Choose the right built-in agent, customize it safely, or create a focused agent for work you repeat.
- [Workflows](https://clarvis.dev/guide/workflows.md): Turn repeatable multi-agent work into a reviewed sequence of focused rounds with explicit inputs and bounded fan-out.

## Extend Clarvis

- [Skills](https://clarvis.dev/guide/skills.md): Package repeatable instructions and supporting resources into commands that users and agents can load only when needed.
- [Hooks](https://clarvis.dev/guide/hooks.md): Run your own checks at Clarvis lifecycle events, and decide exactly which plugin hooks may run.
- [MCP servers](https://clarvis.dev/guide/mcp-servers.md): Connect local or remote Model Context Protocol servers and make their tools, prompts, and resources available to your agents.
- [Plugins](https://clarvis.dev/guide/plugins.md): Bundle agents, skills, MCP servers, hooks, and optional capability services into one extension that remains inactive until you enable it.
- [Marketplaces](https://clarvis.dev/guide/marketplaces.md): Publish and browse curated plugin listings without treating a catalog as permission to run code.

## Reference

- [Configuration](https://clarvis.dev/reference/configuration.md): Understand where Clarvis reads user-owned configuration, how scopes combine, and which settings belong in each file.
- [Commands](https://clarvis.dev/reference/commands.md): Find the stable CLI flags, interactive slash commands, and keyboard routes used in ordinary Clarvis operation.
- [Extensions reference](https://clarvis.dev/reference/extensions.md): Exact locations, configuration fields, namespaces, and activation gates for hooks, MCP servers, skills, plugins, and marketplaces.

## Concepts

- [How Clarvis works](https://clarvis.dev/explanation/how-clarvis-works.md): Understand the few objects Clarvis uses so you can choose the right agent, follow delegated work, and keep long-running projects organized.
- [Scopes and workspace trust](https://clarvis.dev/explanation/scopes-and-trust.md): Decide whether a preference belongs to you or one project, and review repository-controlled code before Clarvis lets it run.

## Operations and security

- [Troubleshooting](https://clarvis.dev/operations/troubleshooting.md): Diagnose readiness, reconnect stale services, recover from memory pressure, and collect bounded diagnostics without guessing which file to edit.
- [Terminal compatibility and accessibility](https://clarvis.dev/terminal-compatibility.md): Understand Clarvis terminal requirements, compatibility controls, and the current accessibility boundaries.
- [Security](https://clarvis.dev/operations/security.md): Keep credentials out of project content, separate command review from process containment, and treat every third-party extension as code you chose to run.

## Complete export

- [All Clarvis documentation](https://clarvis.dev/llms-full.txt): A single consolidated Markdown export for offline retrieval.

## Project

- [Source repository](https://github.com/getclarvis/clarvis): Clarvis source code, releases, issues, and license.