RooCodeInc · brunobergher · May 12, 2026
@@ -1,2 +1 @@
-POSTHOG_API_KEY=your_posthog_api_key
-INTERCOM_APP_ID=your_intercom_app_id
+# No environment variables are required for local docs development.
@@ -36,5 +36,3 @@ jobs:
 
     - name: Build site
       run: pnpm run build
-      env:
-        POSTHOG_API_KEY: ${{ secrets.POSTHOG_API_KEY }}
@@ -6,6 +6,6 @@ This file provides guidance to agents when working with code in this repository.
 - Sitemap uses custom plugin ([docusaurus.config.ts](docusaurus.config.ts:94-107)) filtering "/page/" URLs; preset sitemap disabled. Don't re-enable preset.
 - Redirects required when moving docs ([docusaurus.config.ts](docusaurus.config.ts:109-271)); onBrokenLinks won't fail builds but creates poor UX.
 - Constants centralized in [src/constants.ts](src/constants.ts) for all external URLs. Never hardcode URLs in config or components.
-- PostHog analytics conditionally loaded based on env var ([docusaurus.config.ts](docusaurus.config.ts:84-93)); architectural decision for privacy/GDPR.
+- No analytics or support-chat runtime providers are configured; keep the site focused on open-source documentation surfaces.
 - Search uses local plugin (@easyops-cn/docusaurus-search-local) with docsRouteBasePath "/" - maintain consistency.
-- No blog feature (disabled in preset); this is pure documentation site, not a content platform.
+- No blog feature (disabled in preset); this is pure documentation site, not a content platform.
@@ -9,6 +9,6 @@ This file provides guidance to agents when working with code in this repository.
 - When moving docs, add redirects in [docusaurus.config.ts](docusaurus.config.ts:109); onBrokenLinks is warn ([docusaurus.config.ts](docusaurus.config.ts:40)) so missing redirects won’t fail CI.
 - Internal doc links must be absolute and extensionless per [.roorules](.roorules) (e.g., /basic-usage/how-tools-work).
 - Images in docs must use HTML tags: <img src="/img/...png" alt="..." width="600" /> (Markdown image syntax is disallowed).
-- Builds preload dotenv; PostHog plugin loads only when POSTHOG_API_KEY is set. Use [.env.example](.env.example). See [docusaurus.config.ts](docusaurus.config.ts:83) and [package.json](package.json:7).
+- Builds preload dotenv. No environment variables are required by default; use [.env.example](.env.example) only when adding local-only settings. See [docusaurus.config.ts](docusaurus.config.ts) and [package.json](package.json).
 - Use Node 20 locally to match CI ([.github/workflows/docusaurus-build.yml](.github/workflows/docusaurus-build.yml:23)); package engines allow >=18.
-- “Tutorial Videos” sidebar is driven by [docs/tutorial-videos.json](docs/tutorial-videos.json); titles truncated to 40 chars in [sidebars.ts](sidebars.ts:5).
+- “Tutorial Videos” sidebar is driven by [docs/tutorial-videos.json](docs/tutorial-videos.json); titles truncated to 40 chars in [sidebars.ts](sidebars.ts:5).
@@ -3,8 +3,8 @@
 This file provides guidance to agents when working with code in this repository.
 
 - CI only builds, doesn't lint/check-types ([.github/workflows/docusaurus-build.yml](.github/workflows/docusaurus-build.yml:29)). Run "pnpm run lint" and "pnpm run check-types" locally before pushing.
-- PostHog analytics silently skips if POSTHOG_API_KEY missing. Check [.env](.env) exists with key from [.env.example](.env.example).
+- No environment variables are required by default. Use [.env.example](.env.example) only when debugging local-only settings.
 - No test framework configured. Scripts exist in package.json but no test files or jest/vitest config present.
 - Docusaurus dev server at localhost:3000. Hot reload may fail for theme swizzled components; restart if changes don't appear.
 - onBrokenLinks set to "warn" ([docusaurus.config.ts](docusaurus.config.ts:40)) - broken links won't fail build but check console warnings.
-- Node version mismatch (engines >=18, CI uses 20) can cause subtle differences. Use Node 20 locally.
+- Node version mismatch (engines >=18, CI uses 20) can cause subtle differences. Use Node 20 locally.
@@ -4,7 +4,7 @@ This file provides guidance to agents when working with code in this repository.
 
 Non-obvious, project-specific rules:
 
-- Start/build preload dotenv; analytics plugin is enabled only when POSTHOG_API_KEY is set. Use .env based on [.env.example](.env.example). CI injects it in [.github/workflows/docusaurus-build.yml](.github/workflows/docusaurus-build.yml). See [package.json](package.json) and [docusaurus.config.ts](docusaurus.config.ts).
+- Start/build preload dotenv. No environment variables are required by default; use [.env.example](.env.example) only when adding local-only settings. See [package.json](package.json) and [docusaurus.config.ts](docusaurus.config.ts).
 - Use Node 20 locally to match CI; engines allow >=18 but CI runs 20.
 - Linting targets only /src; docs content is not linted. Use "pnpm run lint:unused" to enforce unused import removal. Type checking uses "tsc" only (no emit); [tsconfig.json](tsconfig.json) is editor-focused.
 - When moving/renaming docs, you must add an explicit redirect in [docusaurus.config.ts](docusaurus.config.ts) under plugin-client-redirects. This is required by [.roorules](.roorules).
@@ -16,4 +16,4 @@ Non-obvious, project-specific rules:
 - Navigation/footer links are centralized in [src/constants.ts](src/constants.ts) and consumed by [docusaurus.config.ts](docusaurus.config.ts). Update constants rather than hardcoding URLs.
 - Legacy [Rakefile](Rakefile) is unrelated (Jekyll). Do not use it; all builds run through Docusaurus scripts.
 - For structural/formatting research, use Context7 MCP (ID "/facebook/docusaurus") via [.roo/mcp.json](.roo/mcp.json); see [.roorules](.roorules).
-- CI runs build and check-types (no lint). Run lint locally to catch issues CI won't.
+- CI runs build and check-types (no lint). Run lint locally to catch issues CI won't.
@@ -47,7 +47,6 @@ This tool marks the end of a task by presenting a final summary of what was acco
 - Optionally demonstrates results through command execution
 - Enables user feedback for further refinements
 - Displays results in a special UI format distinct from regular messages
-- Captures task completion telemetry for system analytics
 - Maintains a structured conversation flow by providing checkpoints
 - Supports subtask completion within larger workflows
 - Ensures users receive a clear summary of what was done

@@ -137,13 +137,12 @@ Creating a new JSON configuration file:
   },
   "features": {
     "darkMode": true,
-    "notifications": true,
-    "analytics": false
+    "notifications": true
   },
   "version": "1.0.0"
 }
 </content>
-<line_count>14</line_count>
+<line_count>13</line_count>
 </write_to_file>
 ```
 

@@ -22,7 +22,7 @@ This page answers some common questions about Roo Code.
 
 ### What is Roo Code?
 
-Roo Code is an AI-powered suite of coding products designed to take full advantage of the most advanced large-language models.
+Roo Code is an open-source AI coding agent for VS Code designed to take full advantage of advanced large-language models.
 
 ### How does Roo Code work?
 
@@ -33,7 +33,7 @@ Roo Code uses large language models (LLMs) to understand your requests and trans
 - Perform web browsing (if enabled)
 - Use external tools via the Model Context Protocol (MCP)
 
-You interact with Roo Code through either a chat interface (in the Extension) or through different channels (in the Cloud).
+You interact with Roo Code through a chat interface in the extension.
 
 ### What can Roo Code do?
 
@@ -50,10 +50,9 @@ Roo Code can help with a variety of coding tasks, including:
 
 ### Is Roo Code free to use?
 
-The Roo Code Extension is free and [open-source](https://github.com/RooCodeInc/Roo-Code/).
-The Roo Code Cloud platform is also free, but some functionality is paid.
+The Roo Code extension is free and [open-source](https://github.com/RooCodeInc/Roo-Code/).
 
-In both cases, Roo Code relies on external LLM inference providers (like [Anthropic](providers/anthropic), [OpenAI](providers/openai), [OpenRouter](providers/openrouter), [Requesty](providers/requesty), etc.) for its AI capabilities. We also offer [Roo Code Router](/roo-code-router/overview) as an alternative.
+Roo Code relies on external LLM inference providers (like [Anthropic](providers/anthropic), [OpenAI](providers/openai), [OpenRouter](providers/openrouter), [Requesty](providers/requesty), etc.) for its AI capabilities.
 These providers typically charge for API usage based on the number of tokens processed.  You will need to create an account and obtain an API key from your chosen provider. Learn more [about providers and how to set them up](/providers/) for details.
 
 ### What are the risks of using Roo Code?
@@ -80,8 +79,6 @@ See the [full list here](/providers/).
 
 Each API provider has its own process for obtaining an API key.  See the [Setting Up Your First AI Provider](/getting-started/connecting-api-provider) for links to the relevant documentation for each provider.
 
-If you use the [Roo Code Router](/roo-code-router/overview), you don't need an API key.
-
 ### Can I use Roo Code with local models?
 Yes, Roo Code supports running models locally using [Ollama](/providers/ollama) and [LM Studio](/providers/lmstudio).  See [Using Local Models](/advanced-usage/local-models) for instructions.
 

@@ -18,9 +18,7 @@ Roo Code needs an inference provider to access the LLM models that make it work.
 
 A great model to start is **Claude Sonnet 4.5**, which offers a lot of power at a reasonable price point. To get it going, choose a provider:
 
-- **Roo Code Router (Recommended):** the easiest way to access multiple providers tested with Roo Code. No API needed. To get started, just choose Roo Code Router as a provider and follow instructions. [Learn more about it](/roo-code-router/overview).
-
-- **OpenRouter:** Provides access to multiple AI models from different labs through a single API key. Great for flexibility and getting started reasonably fast. To get an API key, [follow these instructions <LucideIcon name="ArrowRight" />](/providers/openrouter#getting-an-api-key)
+- **OpenRouter (Recommended):** Provides access to multiple AI models from different labs through a single API key. Great for flexibility and getting started reasonably fast. To get an API key, [follow these instructions <LucideIcon name="ArrowRight" />](/providers/openrouter#getting-an-api-key)
 
 - **Anthropic:** Direct access to the Claude family of models. Requires API access approval and may have [rate limits depending on your tier](https://docs.anthropic.com/en/api/rate-limits#requirements-to-advance-tier).  To get an API key, [follow these instructions <LucideIcon name="ArrowRight" />](/providers/anthropic#getting-an-api-key)
 
@@ -38,8 +36,7 @@ You can choose other models, but that introduces complexity. Different models va
 
 1. Open the Roo Code panel by clicking the Roo Code icon (<KangarooIcon />) in the VS Code Activity Bar
 2. In the welcome screen, choose your LLM provider.
-3. If you choose the Roo Code Router, just connect your account and choose `anthropic/claude-sonnet-4-5` from the dropdown. You're set.
-4. If you chose another provider, paste the API key you copied from it into the right field and continue
-5. Select your model (it should be called `claude-sonnet-4-5` or `anthropic/claude-sonnet-4-5`) and complete the process.
+3. Paste the API key you copied from your provider into the right field and continue.
+4. Select your model (it should be called `claude-sonnet-4-5` or `anthropic/claude-sonnet-4-5`) and complete the process.
 
 Now you can start coding!
@@ -13,19 +13,11 @@ keywords:
 
 # Roo Code Docs
 
-:::danger Sunsetting Roo Code on May 15, 2026
-**All Roo Code products (Extension, Cloud, and Router) will be shut down on May 15, 2026.** We will refund any unused balances. For alternatives, we recommend [Cline](https://github.com/cline/cline) (open-source extension) or [roomote.dev](https://roomote.dev) (our next project). For billing questions after May 15, contact [billing@roocode.com](mailto:billing@roocode.com). [Read the full details](/sunset).
-:::
-
-Roo Code is an AI-coding suite of products designed to take full advantage of the most advanced large-language models and change how you create software at a fundamental level. It's unapologetically powerful and customizable, and it costs more to run than the alternatives because it uses frontier models with actual file system access, terminal control, and multi-step workflows.
+Roo Code is an open-source AI coding agent for VS Code designed to take full advantage of advanced large-language models and change how you create software at a fundamental level. It's powerful, customizable, and model-agnostic, with actual file system access, terminal control, and multi-step workflows.
 
 Roo's approach is to trade tokens for quality. If you want the best and most effective AI coding experience available, this is it.
 
-Roo Code is available in two ways:
-- [<LucideIcon name="Puzzle" style={{ width: "14px", display: "inline-block", verticalAlign: "middle" }} /> Roo Code VS Code Extension](/getting-started/installing) works locally in your VS Code IDE and gives you full control over your coding experience. It's great for deep or highly iterative solo development work. It's highly customizable and fits into your existing workflow.
-- [<LucideIcon name="Cloud" style={{ width: "14px", display: "inline-block", verticalAlign: "middle" }} /> Roo Code Cloud Agents](/roo-code-cloud/cloud-agents) are an autonomous AI development team working 24/7, as a team, collaborating with you and other co-workers on a wide range of tasks. It meets you where you do work (Slack, Github, etc) and allows for large parts of software development to happen entirely away from an IDE.
-
-The two options are highly complementary and built on the same shared foundation, achieving similarly excellent results.
+The Roo Code VS Code extension works locally in your IDE and gives you full control over your coding experience. It's great for deep or highly iterative development work, and it fits into your existing workflow.
 
 ## What can Roo Code do for you
 
@@ -42,13 +34,13 @@ The most successful Roo users tend to embrace the following ideas:
 
 - **Leverage model agnosticism:** Roo isn't an LLM model, it needs an [LLM provider](/providers) to work. But it's compatible with dozens of providers and hundreds of models, so you're free to experiment, optimize and switch around, by design. No lock-ins in a world where "the best model" changes every other week.
 - **Don't skimp on tokens:** expensive state-of-the-art models with lots of tokens will almost always beat cheap models using few tokens. Pay attention to your budget, but embrace the fact that tokens are cheaper than developer time.
-- **Trust roles:** The extension's customizable Modes (Architect, Code, etc) and the pre-defined Cloud Agent Types (Planner, Coder, etc) keep the models on track and focused on the task at hand, without stepping outside of their boundaries. Lean on them depending on the task you're working on.
-- **Be ambitious:** Roo works best the more you trust it. Out-of-the-box, the Extension will ask for your permission to everything, but as you become more comfortable and enable [Auto-Approve](/features/auto-approving-actions), you'll see it work for long periods on its own without getting stuck or making bad decisions. Once ready, throw large, complex projects at the [Orchestrator](/basic-usage/using-modes#orchestrator-mode-aka-boomerang-mode), who will coordinate tasks across Modes. In the Cloud, we handle all of this for you.
+- **Trust roles:** The extension's customizable Modes (Architect, Code, etc) keep the models on track and focused on the task at hand, without stepping outside of their boundaries. Lean on them depending on the task you're working on.
+- **Be ambitious:** Roo works best the more you trust it. Out-of-the-box, the extension will ask for your permission to do everything, but as you become more comfortable and enable [Auto-Approve](/features/auto-approving-actions), you'll see it work for long periods on its own without getting stuck or making bad decisions. Once ready, throw large, complex projects at the [Orchestrator](/basic-usage/using-modes#orchestrator-mode-aka-boomerang-mode), who will coordinate tasks across Modes.
 
 ## Getting Started
 
-- [With the Extension](/getting-started/installing)
-- [With Roo Code Cloud](/roo-code-cloud/overview)
+- [Install the Extension](/getting-started/installing)
+- [Connect an LLM Provider](/getting-started/connecting-api-provider)
 
 
 ## Tutorial & Feature Videos