> ## Documentation Index
> Fetch the complete documentation index at: https://docs.symbioticsec.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# IDE Extension

> Use the IDE extension to get AI assistance directly in your editor

The Symbiotic Code IDE extension brings the terminal coding agent into VS Code-based editors. It provides AI assistance for code generation, debugging, refactoring, and repository exploration. The CLI and IDE extension share the same configuration, agents, permissions, and security modes.

## Supported IDEs

Symbiotic Code supports all IDEs based on VSCode like VSCode, VSCodium, Cursor, Windsurf, Kiro.

## Key features

### Interactive Agent Modes

In the agent sidebar, you can select the most appropriate execution mode for your current task:

* **Build Mode:** Full access to edit files, execute commands, and run tools in a secure environment. Best for active development.
* **Plan Mode:** A read-only analysis agent that inspects codebases, answers architectural questions, and drafts plans. It will always ask for confirmation before making edits or running commands.
* **Debug Mode:** A specialized agent focused on identifying, explaining, and resolving code errors, test failures, or runtime bugs.

You can also create custom modes in **Settings > Custom Modes** by defining a unique name and writing a custom system prompt to tailor the agent's behavior to your team's specific guidelines or coding style.

### Security modes and guardrails

Security modes control how much security workflow runs during a session:

| Security Level         | Description                                                                          | Recommended For                                                          |
| :--------------------- | :----------------------------------------------------------------------------------- | :----------------------------------------------------------------------- |
| **Permissive**         | Minimal guardrails with low-latency checks.                                          | Writing documentation, exploration, and non-sensitive tasks.             |
| **Balanced** (Default) | Context-aware checks that automatically balance developer speed and safety.          | Daily development, refactoring, and general engineering.                 |
| **Strict**             | Comprehensive, maximum-security audits on all inputs, outputs, and terminal actions. | Sensitive codebases, credential management, and high-audit environments. |

Organization and repository guardrails are separate Markdown instructions. Symbiotic Code fetches active guardrails from the backend on startup and refreshes them periodically. See [AI Guardrails](/flow/admins/guardrails).

### Flexible Models & Providers

Configure which models power your development workflow:

* **Symbiotic Tier Models:** Use curated models optimized for security and code generation. Configure variants depending on whether you need high speed or complex reasoning.
* **Default & Mode-Specific Configs:** Navigate to **Settings > Models** to set your default assistant model or assign specific models to different modes (e.g., a lightweight model for Build, a reasoning model for Plan).
* **External Providers:** Seamlessly connect to external AI services like **Anthropic**, **OpenRouter**, or local models running on **Ollama** by configuring your credentials in **Settings > Provider**.

## Getting Started

1. **Install Symbiotic Code** with npm, pnpm, yarn, Bun, or the installer. The same CLI can run in an IDE terminal.
2. **Install the IDE extension** from the VSIX supplied by your Symbiotic administrator or portal, if your team uses the editor integration.
3. **Sign In & Authenticate** with your Symbiotic Security account to connect to the platform.

## Settings

Access the Settings panel in the IDE extension sidebar to customize your Symbiotic Code experience. Settings are organized into the following categories:

### Models

Configure which AI models power your development workflow:

* **Default Model:** The primary model used for conversations. The available default depends on the configured provider.
* **Small Model:** A lightweight model optimized for quick tasks like title generation, commit message generation, and prompt enhancement. Automatically configured.
* **Model per Mode:** Assign specific models to different agent modes:
  * **Build:** Configure a model for active development (defaults to Default Model)
  * **Plan:** Configure a model for analysis and planning (defaults to Default Model)
  * **Debug:** Configure a model for debugging workflows (defaults to Default Model)
  * **Ask:** Configure a model for read-only queries (defaults to Default Model)

### Providers

Manage and configure your AI model providers:

**Primary Provider:**

* **Symbiotic:** The default managed provider using Symbiotic's curated models. Authenticated via the Symbiotic gateway.

**Connected Providers:**

* View all connected third-party providers (e.g., Ollama Cloud, OpenAI, GitHub Copilot)
* Disconnect providers with a single click
* Manage API credentials for each provider

**Popular Providers:**

* **Anthropic:** Direct access to Claude models, including Pro and Max variants
* **Google:** Gemini models for fast, structured responses
* **OpenRouter:** Access all supported models from one unified provider
* **Vercel AI Gateway:** Unified access to AI models with smart routing

Connect any provider by clicking **+ Connect** and entering your API credentials.

### Agent Behaviour

Customize how agents behave and operate with multiple configuration tabs:

**Agents Tab:**
View all available agents and their descriptions:

* **ask:** Get answers and explanations without making changes to the codebase
* **build:** The default agent with all tools enabled for full development access
* **debug:** Diagnose and fix software issues with systematic debugging methodology
* **explore:** Fast agent specialized for exploring codebases and finding files by patterns
* **general:** General-purpose agent for researching complex questions and executing multi-step tasks
* **plan:** Plan mode that disallows all edit tools
* **triage:** Security engineer specialist for triaging static analysis findings

Create new custom modes with **Create New Mode** or import existing agents.

**MCP Servers, Rules, Workflows, Skills Tabs:**
Configure additional agent behavior through Model Context Protocol servers, custom rules, workflow definitions, and skill configurations.

### Auto-Approve

Define how tools are allowed to run with `permission` rules. Permissions resolve to `allow`, `ask`, or `deny` and can match tool inputs with patterns:

**External Directory:**

* **All paths (\*):** Set to "Ask" by default to be prompted before accessing files outside the workspace

**Bash:**

* **Bash:** Configure command patterns such as `git *`, `npm *`, or `rm *` explicitly.
* **Exceptions:** Define specific commands with custom approval settings:
  * `grep *` — Set to "Allow"
  * `git push` — Set to "Ask"
  * `rm *` — Set to "Ask"

**Read:**

* **Environment files:** `*.env` and `*.env.*` require confirmation by default; `*.env.example` is allowed.

Add additional paths or commands with **+ Add path** or **+ Add command**.