# Caudra

> User documentation for the Caudra terminal coding agent.

Full documentation: https://caudra.ai/llms-full.txt

- [Overview](https://caudra.ai/docs/index.md): User documentation for the Caudra terminal coding agent.

## Getting Started

- [Quick Start](https://caudra.ai/docs/quick-start/index.md): Install, connect a provider, first session.
- [Configuration](https://caudra.ai/docs/configuration/index.md): caudra.toml settings, experimental features, separate profile files, and directory layout.

## Guides

- [Skills](https://caudra.ai/docs/skills/index.md): Write Markdown playbooks the agent loads on demand.
- [Remote workspaces](https://caudra.ai/docs/remote-workspaces/index.md): Connect to Workcell with explicit workspace identity and remote execution (experimental).
- [Managed sandboxes](https://caudra.ai/docs/sandboxes/index.md): Manage VM profiles, images, network policy and reviewed file transfers (experimental).
- [Sessions, forks, and revert](https://caudra.ai/docs/sessions/index.md): Fork conversation points and revert chat or file changes.
- [Cross-session messaging](https://caudra.ai/docs/messaging/index.md): Let live sessions and scripts on one machine message each other (experimental).
- [Worktrees](https://caudra.ai/docs/worktrees/index.md): Create, open and remove git worktrees, with or without Herdr.
- [Review](https://caudra.ai/docs/review/index.md): Mark passages of a reply and send notes on them back.
- [Workbench](https://caudra.ai/docs/workbench/index.md): File explorer, editor, source control, and search beside the transcript.
- [System Prompt Profiles](https://caudra.ai/docs/system-prompts/index.md): Create named prompt profiles without copying Caudra's dynamic prompt.
- [Writing caudra plugins](https://caudra.ai/docs/plugins/index.md): Add your own tools and commands in Lua, or let the agent write them (experimental).
- [Workflows](https://caudra.ai/docs/workflows/index.md): Durable multi-agent scripts that pause, resume, and report back (experimental).
- [Automations](https://caudra.ai/docs/automations/index.md): Rhai scripts that react to session events and schedules (experimental).
- [Headless Mode](https://caudra.ai/docs/headless/index.md): --print for scripts and CI. Drop-in Claude Code compatible.
- [ACP (Agent Client Protocol)](https://caudra.ai/docs/acp/index.md): Drive Caudra from your editor, like Zed, over the Agent Client Protocol.

## Concepts

- [Token Economy](https://caudra.ai/docs/token-economy/index.md): Where tokens go in an agent loop, and every trick Caudra uses to spend fewer of them.
- [Context](https://caudra.ai/docs/context/index.md): What enters the model's context and when, and where to put project knowledge.
- [Changes from Maki](https://caudra.ai/docs/changes-from-maki/index.md): What changed after the fork point, and why each change exists.
- [Queue and Steering](https://caudra.ai/docs/queue/index.md): Send work next, guide the current run, or stop and replace it.

## Reference

- [Tools](https://caudra.ai/docs/tools/index.md): Every built-in tool and its parameters.
- [Providers](https://caudra.ai/docs/providers/index.md): Model catalogs, env vars, providers.toml, model jobs.
- [Permissions](https://caudra.ai/docs/permissions/index.md): What runs freely, what asks first, TOML rules.
- [Notifications](https://caudra.ai/docs/notifications/index.md): Know when a session finishes or needs your input.
- [MCP (Model Context Protocol)](https://caudra.ai/docs/mcp/index.md): External tool servers over stdio or HTTP.
- [Commands](https://caudra.ai/docs/commands/index.md): The / palette, sessions, toggles, custom commands.
- [Keybindings](https://caudra.ai/docs/keybindings/index.md): Defaults, precedence, rebinding from Lua.
- [Markdown](https://caudra.ai/docs/markdown/index.md): How replies render: tables, code, maths and Mermaid flowcharts.
- [Lua API](https://caudra.ai/docs/lua-api/index.md): The plugin surface, mirrored from Neovim (experimental).
- [CLI](https://caudra.ai/docs/cli/index.md): Flags and subcommands (auth, models, acp, prompt, ...).
- [Reference configs](https://caudra.ai/docs/reference-configs/index.md): Every TOML config file in full, with each key commented out and described.
- [Logging](https://caudra.ai/docs/logging/index.md): The structured log every run writes, and the two ways to read it.
- [Telemetry](https://caudra.ai/docs/telemetry/index.md): Opt-in OpenTelemetry metrics and events, to a collector you run.