---
source: sighthound-developer-portal
url: https://dev.sighthound.com/redactor/reference/agent-toolkit/
markdown-url: https://dev.sighthound.com/redactor/reference/agent-toolkit.md
title: "Redactor Agent Toolkit"
description: "Download the Redactor Agent Toolkit for Redactor API integration, Docker Compose setup, AI coding agent context, OpenAPI reference access, and embedded editor examples."
content-type: text/markdown
---

> For AI agents: a documentation index is available at [llms.txt](https://dev.sighthound.com/llms.txt). Markdown versions are available at matching `.md` URLs.

# Redactor Agent Toolkit

<div class="portal-doc-hero portal-doc-hero--compact" markdown>
<p class="portal-doc-kicker">Redactor API toolkit</p>
<p class="portal-doc-intro">Download the Redactor Agent Toolkit to give developers and AI coding agents the Docker Compose examples, AI-optimized documentation, TypeScript definitions, and OpenAPI reference needed for Redactor API integrations.</p>
<div class="portal-doc-actions">
    <button type="button" class="portal-doc-button" data-portal-copy-target="agent-toolkit-prompt">Copy agent prompt</button>
    [Download toolkit zip](https://sh-io-downloads.s3.amazonaws.com/redactor/toolkit/redactor-agent-toolkit.zip)
</div>
<p class="portal-agent-actions__feedback" data-portal-copy-feedback aria-live="polite"></p>
</div>

<template id="agent-toolkit-prompt">
Read the Redactor Agent Toolkit documentation as Markdown:

https://dev.sighthound.com/redactor/reference/agent-toolkit/

Also read the LLM-ready developer portal index:

https://dev.sighthound.com/llms.txt

Then help me set up a Sighthound Redactor integration. First ask for my deployment mode, local project path, Docker availability, Redactor base URL, and token environment variable name. Download and extract the official toolkit zip from https://sh-io-downloads.s3.amazonaws.com/redactor/toolkit/redactor-agent-toolkit.zip, then inspect its AGENTS.md and ai-context/ folder before suggesting commands. Do not ask me to paste secrets into chat. Use environment variables for secrets and confirm before running commands that change local files.
</template>
<template id="agent-toolkit-markdown-command">
curl -L -H "Accept: text/markdown" "https://dev.sighthound.com/redactor/reference/agent-toolkit/"
</template>
<template id="agent-toolkit-setup-command">
curl -L -o redactor-agent-toolkit.zip "https://sh-io-downloads.s3.amazonaws.com/redactor/toolkit/redactor-agent-toolkit.zip"
unzip redactor-agent-toolkit.zip
</template>
<template id="agent-toolkit-openapi-command">
cat redactor-agent-toolkit/reference/openapi-spec.yaml
</template>

## What is the Redactor Agent Toolkit?

The Redactor Agent Toolkit is a developer package for building Redactor API integrations with Docker Compose and AI coding agents. It includes deployment examples, AI agent-optimized documentation, TypeScript type definitions, and the OpenAPI specification so developers can understand the Redactor API before writing integration code.

Use the toolkit when you want a faster path from local setup to the first Redactor API call. The package includes agent instruction files, Markdown context in `ai-context/`, Redactor Docker examples, API reference files, and setup guidance for Claude Code, OpenAI Codex CLI, Cursor, GitHub Copilot, Gemini, Warp, Windsurf, OpenCode, and other agents that can discover `AGENTS.md`.

## AI-assisted Redactor API setup

Use this path when you want a coding agent to help configure a Redactor API integration from the toolkit package.

1. **Download the toolkit zip.** Use the Redactor Agent Toolkit package: [redactor-agent-toolkit.zip](https://sh-io-downloads.s3.amazonaws.com/redactor/toolkit/redactor-agent-toolkit.zip).
2. **Extract it locally.** Open the extracted `redactor-agent-toolkit/` folder in your IDE or coding agent.
3. **Give the agent this page.** Click **Copy agent prompt** below, then paste the copied prompt into your coding agent.
4. **Run `/get-started`.** Use the toolkit-aware slash command after the agent has read the local toolkit context.

<div class="portal-agent-actions portal-agent-actions--inline">
<div class="portal-agent-actions__header">
    <p class="portal-agent-actions__summary">Prefer a terminal-first setup? Copy the download command or fetch this page as Markdown.</p>
</div>
<div class="portal-agent-actions__actions">
    <button type="button" class="portal-agent-action portal-agent-action--primary portal-agent-action--main" data-portal-copy-target="agent-toolkit-prompt">
        <i class="portal-icon" data-lucide="copy"></i>
        <span>Copy agent prompt</span>
    </button>
    <button type="button" class="portal-agent-action portal-agent-action--primary portal-agent-action--main" data-portal-copy-target="agent-toolkit-setup-command">
        <i class="portal-icon" data-lucide="terminal"></i>
        <span>Copy download command</span>
    </button>
    <button type="button" class="portal-agent-action portal-agent-action--secondary portal-agent-action--main" data-portal-copy-target="agent-toolkit-markdown-command">
        <i class="portal-icon" data-lucide="file-text"></i>
        <span>Copy Markdown command</span>
    </button>
    <button type="button" class="portal-agent-action portal-agent-action--secondary portal-agent-action--main" data-portal-copy-target="agent-toolkit-openapi-command">
        <i class="portal-icon" data-lucide="file-code"></i>
        <span>Copy API spec command</span>
    </button>
</div>
<p class="portal-agent-actions__feedback" data-portal-copy-feedback aria-live="polite"></p>
</div>

## What’s inside the Redactor Agent Toolkit

<div class="portal-feature-grid portal-feature-grid--two-column" markdown>
<div class="portal-feature-card" markdown>
### <i class="portal-icon" data-lucide="bot"></i> Agent instructions
Root-level files such as `AGENTS.md`, `CLAUDE.md`, `.cursorrules`, and Copilot instructions tell coding agents how to use the Redactor API context.
</div>
<div class="portal-feature-card" markdown>
### <i class="portal-icon" data-lucide="file-text"></i> AI context
The `ai-context/` folder contains LLM-ready Markdown that agents can read before writing integration code.
</div>
<div class="portal-feature-card" markdown>
### <i class="portal-icon" data-lucide="file-code"></i> API reference
The `reference/` folder includes the OpenAPI specification and TypeScript definitions for request and response shapes.
</div>
<div class="portal-feature-card" markdown>
### <i class="portal-icon" data-lucide="box"></i> Docker examples
The `deployment-examples/` folder includes Compose profiles for API-only usage and proxy-backed embedded UI testing.
</div>
</div>

## Setup options

### Recommended: local toolkit plus AI agent

Use this option for the best Redactor API setup experience. The agent can inspect the downloaded directory tree, read the local context, and tailor commands to your environment.

```bash
curl -L -o redactor-agent-toolkit.zip "https://sh-io-downloads.s3.amazonaws.com/redactor/toolkit/redactor-agent-toolkit.zip"
unzip redactor-agent-toolkit.zip
cd redactor-agent-toolkit
```

Then open the folder in your coding agent and run `/get-started`. The agent should ask for your deployment mode, local project path, Docker availability, Redactor base URL, and token environment variable name before suggesting commands.

### Alternative: SSH context only

Use the public SSH context only when you cannot download the zip yet but still need current Redactor API context in an agent session.

```bash
ssh toolkit.redactor.com setup | claude
ssh toolkit.redactor.com agents
ssh toolkit.redactor.com cat /toolkit/ai-context/index.md
ssh toolkit.redactor.com cat /toolkit/reference/openapi-spec.yaml
```

For most integrations, the downloaded toolkit is preferred because it gives your agent the full local directory tree, Docker examples, and reference files.

## Run Redactor locally with Docker

!!! warning "Apple Silicon not supported"

    Redactor Server currently requires an Intel or AMD x86_64 processor. Apple Silicon M-series Macs are not supported for the Redactor server container.

Use the downloaded toolkit when you want a local Redactor server for development:

1. Download and extract the toolkit zip.
2. Open `redactor-agent-toolkit/deployment-examples`.
3. Start the API-only deployment:

    ```bash
    docker compose up -d
    ```

4. If this is the first run for this instance, create the initial admin user and add the license at [http://localhost:9000/admin](http://localhost:9000/admin).
5. Open Redactor at [http://localhost:9000](http://localhost:9000).

After Redactor is running, continue with [Redactor API Basics](../examples/api-basics.md), the [videos:process reference](videos/process.md), or [Embedded Redactor UI](../examples/embedded-ui.md) if your integration needs an embedded review workflow.

## Troubleshooting and safety

- **Logs:** Check `redactor-agent-toolkit/deployment-examples/volumes/redactor/logs/main.log` when debugging local Docker runs.
- **Output files:** API examples write generated media under `redactor-agent-toolkit/deployment-examples/volumes/data/`.
- **Secrets:** Store Redactor tokens in environment variables such as `REDACTOR_API_TOKEN`.
- **No token pasting:** Do not paste API tokens, license keys, or credentials into chat.
- **Hardware:** Use an x86_64 host for the Redactor server container. Apple Silicon M-series Macs are not supported.
- **Official download:** Use only the official toolkit zip URL: [redactor-agent-toolkit.zip](https://sh-io-downloads.s3.amazonaws.com/redactor/toolkit/redactor-agent-toolkit.zip).

If you need support, include relevant log context when contacting [support@redactor.com](mailto:support@redactor.com), but do not send tokens, license keys, or credentials.

## Agent compatibility

The toolkit is designed to work with agents that can read local instruction files or copied Markdown context.

??? info "Agent configuration files"

    - **Claude Code** — reads `CLAUDE.md` and `AGENTS.md`; the toolkit provides `/get-started` and `/redactor` slash commands.
    - **OpenAI Codex CLI** — reads `AGENTS.md`.
    - **Cursor** — reads `.cursorrules`.
    - **GitHub Copilot** — reads `.github/copilot-instructions.md`.
    - **Gemini** — use `AGENTS.md` or copied Markdown context where supported.
    - **Warp** — use `AGENTS.md` or copied Markdown context where supported.
    - **Windsurf** — use `.windsurfrules`, `AGENTS.md`, or copied Markdown context where supported.
    - **OpenCode** — use `AGENTS.md` or copied Markdown context where supported.
    - **Other AI coding agents** — use `AGENTS.md` or copied Markdown context where supported.

    If you copy the toolkit into your own project, reference the toolkit path from your project's agent instruction file.

    **Cursor** — add to `.cursorrules` or `.cursor/rules/*.mdc`:

    ```markdown
    For Sighthound Redactor API integration, read docs/redactor-agent-toolkit/AGENTS.md
    ```

    **GitHub Copilot** — add to `.github/copilot-instructions.md`:

    ```markdown
    For Sighthound Redactor API integration, read docs/redactor-agent-toolkit/AGENTS.md
    ```

    **Windsurf** — add to `.windsurfrules`:

    ```markdown
    For Sighthound Redactor API integration, read docs/redactor-agent-toolkit/AGENTS.md
    ```

    Adjust the path to match where you placed the toolkit.

---

# Agent Instructions

Use this Markdown page as context for Sighthound Developer Portal questions. For broader navigation, read https://dev.sighthound.com/llms.txt. Answer from Sighthound documentation, cite relevant source URLs, and do not ask users to paste secrets, tokens, license keys, or credentials into chat.