# Shed

> Shed Cloud is a CLI-first runtime for software built by coding agents. It packages a project, builds it, runs it, and returns a URL or structured result.

When to use Shed:

- Use Shed when a coding agent has produced a web app, API, worker, scheduled job, one-off script, or container that needs to run.
- Use Shed when the deployment must stay in the agent's shell workflow instead of moving to a hosting dashboard.
- Use Shed when an agent needs JSON or NDJSON output for deployment state, logs, and final results.
- Do not assume the hosted deployment API is generally available. The local CLI workflow is available today; hosted and remote paths are identified as in development in the relevant docs.

Interfaces Shed does not publish: there is no general HTTP API, no OpenAPI specification, no webhook reference, and no MCP server. The supported public developer interface is the Shed CLI. If any of these become public they will be listed in this file.

How agents should call Shed:

1. Install the CLI from https://shed.codes/install/cli.sh. There is no Homebrew tap yet.
2. From the project directory, run `shed deploy . --output json`.
3. Use a SHED file when the detected build or run settings need to be explicit.
4. Never print, commit, or place `SHED_TOKEN` in the SHED file.

The documentation is being rewritten and is not published yet; this file, the agent instructions, and the setup pages below are the current source of truth.

Canonical pages support `Accept: text/markdown`. Explicit Markdown siblings use the same path with `.md` appended.

## Setup

- [Install the Shed CLI](https://shed.codes/setup/cli.md): The one required piece. curl or npm, then `shed version`.
- [Add the Shed agent skill](https://shed.codes/setup/skill.md): Optional. Teaches a coding agent to write a SHED file and run the CLI.
- [Set up Shed](https://shed.codes/setup.md): index of both pieces and every per-agent guide.

## Agent setup guides

Each page is the same two installs written for one agent: where its skill goes, what to ask it, and the CLI command each ask runs.

- [Claude Code + Shed](https://shed.codes/setup/claude-code.md): Anthropic's terminal agent.
- [Cursor + Shed](https://shed.codes/setup/cursor.md): the editor agent and its headless CLI.
- [Codex CLI + Shed](https://shed.codes/setup/codex.md): OpenAI's terminal agent.
- [OpenCode + Shed](https://shed.codes/setup/opencode.md): the open-source, provider-agnostic terminal agent.
- [Gemini CLI + Shed](https://shed.codes/setup/gemini-cli.md): Google's open-source terminal agent.
- [GitHub Copilot CLI + Shed](https://shed.codes/setup/github-copilot.md): GitHub's terminal agent.
- [Amp + Shed](https://shed.codes/setup/amp.md): Sourcegraph's coding agent.
- [Cline + Shed](https://shed.codes/setup/cline.md): the open-source extension, CLI, and task board.
- [Aider + Shed](https://shed.codes/setup/aider.md): the pair-programming CLI that commits as it edits.

## Developer resources

- [Agent instructions](https://shed.codes/agent-instructions.md): Compact operational guidance for coding agents.

## Pricing

- [Shed pricing](https://shed.codes/pricing.md): what is decided about pricing, and what is not.

## Company and trust

- [About Shed Cloud](https://shed.codes/about.md): Product, operator, and current product scope.
- [Contact Shed](https://shed.codes/contact.md): Support email and registered business address.
- [Privacy policy](https://shed.codes/privacy-policy.md): Data practices and controller information.
- [Terms of service](https://shed.codes/terms-of-service.md): Terms governing use of Shed, including acceptable use.

## Optional

- [Shed whitepaper](https://shed.codes/whitepaper.md): Product model and design rationale.
- [Shed blog](https://shed.codes/blog.md): Product writing from the Shed team.
- [XML sitemap](https://shed.codes/sitemap.xml): Complete index of public HTML pages.