Supported AI Agents
Agent families represented in the VibeAround launch and continuation model.
Which coding agents, IM channels, and model providers VibeAround supports, and what each combination can do. Ids in the tables are the values you use in settings.json, slash commands, and CLI flags.
Coding agents
VibeAround drives agents over the Agent Client Protocol (ACP). Every hosted agent supports the full conversation flow: prompting, streaming output, permission requests, and cancellation.
| Agent id | Product | Hosted (IM / web chat) | Session resume | Notes / aliases |
|---|---|---|---|---|
claude | Claude Code | Yes | Yes | claude-code |
codex | Codex CLI | Yes | Yes | openai-codex |
gemini | Gemini CLI | Yes | Yes | gemini-cli |
cursor | Cursor CLI | Yes | Yes | — |
qwen-code | Qwen Code | Yes | Yes | qwen |
opencode | Opencode | Yes | Yes | open-code |
pi | Pi | Yes | Yes | pi-agent, pi-coding-agent |
kiro | Kiro CLI | Yes | No resume template yet | kiro-cli |
claude-desktop | Claude Desktop | Launch target only | — | Opens the desktop app; not an ACP runtime |
codex-desktop | Codex Desktop | Launch target only | — | Opens the desktop app; not an ACP runtime |
"Session resume" means VibeAround can restore the agent's own CLI session across idle shutdowns, daemon restarts, and handovers.
IM channels
Channels are plugins (separate npm packages); the web chat and TUI are built-in surfaces that use the same routing. All channels support the core flow: prompts, streamed replies, slash commands, and permission cards.
| Channel | Kind id | Delivered as | Repository |
|---|---|---|---|
| Telegram | telegram | Plugin | va-plugin-channel-telegram |
| Slack | slack | Plugin | va-plugin-channel-slack |
| Feishu / Lark | feishu | Plugin (interactive cards use the V2 card schema) | va-plugin-channel-feishu |
| Discord | discord | Plugin | va-plugin-channel-discord |
weixin-openclaw-bridge | Plugin | va-plugin-channel-weixin-openclaw-bridge | |
whatsapp | Plugin | va-plugin-channel-whatsapp | |
| DingTalk | dingtalk | Plugin | va-plugin-channel-dingtalk |
| WeCom (企业微信) | wecom | Plugin | va-plugin-channel-wecom |
| QQ Bot | qqbot | Plugin | va-plugin-channel-qqbot |
| Web chat | web | Built-in (dashboard) | — |
| TUI chat | tui | Built-in (vibearound tui) | — |
Per-channel setup pages (platform steps + verified config blocks) live under guides/channels/; configuration lives under channels.<kind> in settings.json.
Model providers (profile catalog)
The built-in catalog ships endpoint and model definitions for these providers. Each endpoint declares which API dialects it serves — the bridge translates whenever your agent's dialect differs (see Local API and bridge).
| Provider id | Label | Upstream dialects available | Notes |
|---|---|---|---|
moonshot | Kimi / Moonshot | openai-chat, anthropic | Global + CN endpoints, Kimi coding plan |
dashscope | Alibaba DashScope | anthropic, openai-responses, openai-chat | CN endpoints, coding/token plans |
deepseek | DeepSeek | anthropic, openai-chat | Dedicated translation adapter |
openrouter | OpenRouter | openai-chat | — |
minimax | MiniMax | anthropic, openai-chat, openai-responses | Global + CN, token plans |
mimo | Xiaomi MiMo | openai-chat, anthropic | Pay-as-you-go + CN token plan |
volcengine | Volcengine | anthropic, openai-responses, openai-chat | Ark API, coding/agent plans |
zai | Z.AI / GLM | openai-chat, anthropic | Global + CN, coding plans |
gemini | Google Gemini / Vertex AI | gemini, openai-chat | Gemini API + Vertex-compatible |
xai | xAI / Grok | openai-responses, openai-chat | — |
nvidia | NVIDIA NIM | openai-chat | — |
azure | Azure OpenAI | openai-responses | Custom deployment fields |
A custom profile (any OpenAI/Anthropic/Gemini-compatible endpoint) is always available beyond the catalog, and the direct profile launches an agent with its own vendor login and no bridge.
Plan-by-plan detail — endpoint groups, regions, base URLs, model sets, and which credential each plan expects — is in the provider endpoints reference.
Platforms
| Platform | Desktop app | Standalone server / CLI |
|---|---|---|
| macOS Apple Silicon | Packaged (DMG) | Yes |
| macOS Intel | Source build | Yes |
| Windows x64 | Packaged (EXE/MSI/portable) | Yes |
| Linux x64 | Packaged (AppImage/deb); terminal launch depends on desktop environment | Yes |
Source anchors: src/resources/agents.json (agent registry — ids, aliases, direct_only, resume templates), src/resources/plugins.json (channel plugin registry — kind ids and repositories), src/resources/profile-catalog/ (provider endpoints and dialects), src/server/src/lib.rs (built-in web/tui channels), README.md (packaging status).
Last verified: v0.7.11
Tunnels & Remote Access
Reach the VibeAround dashboard from anywhere with ngrok, localtunnel, or Cloudflare Tunnel, protected by pairing codes.
VibeAround Architecture
The technical model behind VibeAround's workspace for local AI coding agents, API Bridge, remote entry points, and session coordination.