> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/eugene1g/agent-safehouse/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent Compatibility Overview

> Complete list of coding agents supported by Agent Safehouse with sandboxing status and integration details

Agent Safehouse provides macOS sandbox wrappers for 15 coding agents, constraining their filesystem and network access using Apple's `sandbox-exec` (Seatbelt) framework.

## Supported Agents

<CardGroup cols={3}>
  <Card title="Aider" icon="python" href="/agents/aider">
    Python CLI assistant with terminal UI
  </Card>

  <Card title="Amp" icon="bolt" href="/agents/amp">
    Clipboard-aware coding agent
  </Card>

  <Card title="Auggie" icon="a" href="/agents/auggie">
    Augment Code CLI agent
  </Card>

  <Card title="Claude Code" icon="c" href="/agents/claude-code">
    Anthropic's CLI/TUI agent
  </Card>

  <Card title="Cline" icon="code" href="/agents/cline">
    VS Code extension agent
  </Card>

  <Card title="Codex" icon="cube" href="/agents/codex">
    OpenAI Codex CLI (Rust)
  </Card>

  <Card title="Copilot CLI" icon="github" href="/agents/copilot-cli">
    GitHub Copilot command line
  </Card>

  <Card title="Cursor Agent" icon="cursor" href="/agents/cursor-agent">
    Cursor IDE agent mode
  </Card>

  <Card title="Droid" icon="robot" href="/agents/droid">
    Factory CLI coding agent
  </Card>

  <Card title="Gemini CLI" icon="google" href="/agents/gemini-cli">
    Google's Gemini CLI agent
  </Card>

  <Card title="Goose" icon="goose" href="/agents/goose">
    Block's open-source agent
  </Card>

  <Card title="Kilo Code" icon="k" href="/agents/kilo-code">
    VS Code extension (Cline fork)
  </Card>

  <Card title="OpenCode" icon="code" href="/agents/opencode">
    Open-source CLI/TUI agent
  </Card>

  <Card title="Pi" icon="pi" href="/agents/pi">
    Minimal CLI coding agent
  </Card>
</CardGroup>

## How Agent Profiles Work

Each agent has a dedicated sandbox profile in `profiles/60-agents/<agent>.sb` that grants:

* **Binary paths** -- Where the agent CLI is installed (`~/.local/bin/`, brew, npm global, etc.)
* **Config directories** -- Agent settings, sessions, logs (`~/.agent-name/`)
* **State/cache** -- Session data, model caches, MCP configurations
* **Integrations** -- Keychain access (when needed), MCP paths, hooks directories

Profiles are **layered** on top of base system access. See [Policy Architecture](/advanced/policy-architecture) for the complete policy stack.

## Shell Wrapper Functions

Safehouse provides shell functions for each agent. Add to your `~/.zshrc` or `~/.bashrc`:

```bash theme={null}
# Source the generated shell wrappers
source ~/path/to/safehouse/dist/safehouse.sh

# Now use agents with automatic sandboxing:
aider --help
claude "write a function"
goose
```

Each function wraps the agent binary with `safehouse --` to apply the sandbox policy.

## Sandbox Policy Examples

### Read-Only Workspace (Default)

```bash theme={null}
# Agent can read workspace but not modify files
safehouse -- aider --no-auto-commits
```

### Workspace Write Access

```bash theme={null}
# Allow file modifications in current directory
safehouse --add-dirs=$PWD -- claude "refactor this"
```

### Network Isolation

By default, agents have **no network access** in the sandbox. Enable selectively:

```bash theme={null}
# Enable network for LLM API calls
safehouse --enable=network -- goose
```

### Custom Integrations

Some agents require additional integrations:

```bash theme={null}
# Claude Code with keychain + browser
safehouse --enable=keychain,browser-native-messaging -- claude

# Aider with clipboard (Amp-style)
safehouse --enable=clipboard -- amp
```

## Agent Investigation Reports

Each agent has a detailed investigation report analyzing:

* UI architecture (TUI framework, Electron, web UI)
* Authentication mechanisms (API keys, OAuth, keychain)
* Filesystem access patterns (config, state, temp files)
* Network endpoints (LLM APIs, telemetry, update checks)
* Subprocess execution (shell commands, git, npm, docker)
* Known vulnerabilities and CVEs

See the individual agent pages for complete analysis.

## Adding New Agents

To add support for a new agent:

1. **Create profile** -- `profiles/60-agents/new-agent.sb` with binary and config paths
2. **Test access** -- Use `--explain` to verify granted paths match agent needs
3. **Add dependencies** -- Use `$$require=path/to/integration.sb$$` for keychain, clipboard, etc.
4. **Regenerate dist** -- Run `./scripts/generate-dist.sh` to update packaged artifacts
5. **Test sandbox** -- Verify agent works with minimal policy

See [Contributing](/operations/contributing) for detailed instructions.

## Comparison Matrix

| Agent        | Language   | Keychain | Network Default | Subprocess       | UI         |
| ------------ | ---------- | -------- | --------------- | ---------------- | ---------- |
| Aider        | Python     | No       | Denied          | Shell            | Terminal   |
| Amp          | Bun        | No       | Denied          | Shell            | TUI        |
| Auggie       | Node.js    | No       | Denied          | Shell            | TUI        |
| Claude Code  | Rust       | Yes      | Denied          | Sandboxed        | TUI        |
| Cline        | TypeScript | Yes      | Denied          | VS Code Terminal | Webview    |
| Codex        | Rust       | Yes      | Denied          | Sandboxed        | TUI        |
| Copilot CLI  | Node.js    | Yes      | Denied          | Shell            | Terminal   |
| Cursor Agent | Electron   | Yes      | Denied          | Shell            | Electron   |
| Droid        | Bun        | No       | Denied          | Shell            | TUI        |
| Gemini CLI   | Node.js    | Yes      | Denied          | Shell            | TUI        |
| Goose        | Rust       | Yes      | Denied          | Shell            | TUI        |
| Kilo Code    | TypeScript | Yes      | Denied          | VS Code Terminal | Webview    |
| OpenCode     | Node.js    | No       | Denied          | Shell            | TUI        |
| Pi           | Node.js    | No       | Denied          | Shell            | Custom TUI |

## Special Integration Notes

### VS Code Extensions (Cline, Kilo Code)

VS Code extensions run in the Extension Host process, inheriting VS Code's sandbox. Safehouse **cannot** directly sandbox VS Code extensions. Instead:

* Use `--add-dirs` to restrict workspace access via VS Code's file watchers
* Extensions inherit VS Code's network policy
* Terminal commands spawn through VS Code's Terminal API

For stronger isolation, run VS Code itself in a container.

### Electron Apps (Cursor)

Electron apps have their own Chromium sandbox. Safehouse profiles grant access to:

* App bundle (`/Applications/Cursor.app`)
* Application Support directories
* Temp directories for IPC

Cursor's agent mode runs within the Electron renderer with standard Chromium isolation.

### Rust Agents (Codex, Goose)

Rust agents (Codex, Goose) have **built-in sandboxing**:

* **Codex** uses `sandbox-exec` (macOS Seatbelt) for its own tool calls
* **Goose** has no built-in sandbox but is compiled native code

Safehouse profiles grant access to their binaries and config paths. The agent's own sandboxing is orthogonal.