ch1bug/dsh-mimo-agent-tools
30Last commit Aug 14, 2026
dsh-mimo-agent-tools DSH plugin
mimo-agent-tools is a DSH Cordis plugin that integrates the Xiaomi MiMo API, providing a set of tools for AI agents: web search, multimodal understanding (image, audio, video), speech-to-text, and text-to-speech. It uses a Python driver to handle large multimodal payloads without shell truncation issues.
How to install the dsh-mimo-agent-tools DSH plugin
dsh plugin --profile web add /path/to/dsh-mimo-agent-toolsCopying does not run this command. Review the repository and version before installing the dsh-mimo-agent-tools DSH plugin.
dsh-mimo-agent-tools DSH plugin data source
dsh-mimo-agent-tools DSH plugin snapshot date: Aug 16, 2026
discovered
What the dsh-mimo-agent-tools DSH plugin can do
- Web search via `mimo_search` tool, returning answer with cited sources
- Image understanding via `mimo_vision` – supports local files (auto base64) and public URLs, multi-image
- Audio understanding/transcription via `mimo_audio` – supports wav/mp3/flac/ogg/m4a
- Video understanding via `mimo_video` – supports mp4/webm/mov from public URLs
- Speech-to-text via `mimo_asr` with optional language hint
- Text-to-speech via `mimo_tts` – preset voices or free-form voice design, output to .wav file
Where the dsh-mimo-agent-tools DSH plugin fits
- Enable AI agents to search the web and cite sources directly
- Analyze images, audio, or video content from local files or URLs
- Convert speech to text for transcription or voice commands
- Generate text-to-speech audio with custom voice design
Who the dsh-mimo-agent-tools DSH plugin is for
- DSH users who want to add Xiaomi MiMo AI capabilities to their agents
- Developers building AI assistants that need multimodal understanding and speech features
dsh-mimo-agent-tools DSH plugin limitations
- Requires a MiMo API key; the `web_search` tool also needs the connected-search plugin enabled in the MiMo console
- Depends on Python 3 installed on the host (uses a Python driver)
- TTS with the `voicedesign` model does not accept `audio.voice` field; uses `optimize_text_preview` instead
- ASR model must not receive a `thinking` field; web search costs per keyword round
dsh-mimo-agent-tools DSH plugin: from the repository README
Quoted from the ch1bug/dsh-mimo-agent-tools README, the upstream source of the dsh-mimo-agent-tools DSH plugin. Copyright remains with the original authors.
DSH (DeepSeek Harness) Cordis plugin that turns the **Xiaomi MiMo API** into model tools for an agent: web search, image/audio/video understanding, speech-to-text, and text-to-speech. Backed by the OpenAI-compatible endpoint `https://api.xiaomimimo.com/v1`. ## Tools | Tool | Model | Purpose | |---|---|---| | `mimo_search` | mimo-v2.5-pro | Web search via the native `web_search` tool, returns answer + cited sources | | `mimo_vision` | mimo-v2.5 | Image understanding — local files (auto base64) or public URLs, multi-image | | `mimo_audio` | mimo-v2.5 | Audio understanding / transcription (wav/mp3/flac/ogg/m4a) | | `mimo_video` | mimo-v2.5 | Video understanding from a public URL (mp4/webm/mov) | | `mimo_asr` | mimo-v2.5-asr | Speech-to-text with optional language hint | | `mimo_tts` | mimo-v2.5-tts / -voicedesign | Text-to-speech to a `.wav` file — preset voices or free-form voice design | ## Requirements - A [MiMo API key](https://mimo.mi.com) — the `web_search` tool requires the connected-search plugin enabled in the MiMo console. - `python3` on the host (used by `driver/mimo_driver.py`). - DSH host with the `shell` and `sandboxPolicy` services. ## Configuration The API key
Read the full READMERepository license: MIT
dsh-mimo-agent-tools DSH plugin questions
How do I install the MiMo Agent Tools plugin?
You can install it via the DSH plugin command: 'dsh plugin --profile web add /path/to/dsh-mimo-agent-tools' or 'dsh plugin --profile web add github:you/dsh-mimo-agent-tools'. After installation, copy the Python driver to the default path: 'mkdir -p ~/.local/lib/mimo-agent-tools/driver && cp driver/mimo_driver.py ~/.local/lib/mimo-agent-tools/driver/'. Then restart DSH web; the tools will mount automatically.
How do I configure the API key?
The plugin resolves the API key from the DSH credentials service first (key name XIAOMI_API_KEY, set via the web Models page), then falls back to the environment variable XIAOMI_API_KEY or MIMO_API_KEY. No hardcoded keys are used. You can also set the key in the environment if you prefer.
What tools does this plugin provide?
It provides six tools: mimo_search (web search), mimo_vision (image understanding), mimo_audio (audio understanding), mimo_video (video understanding), mimo_asr (speech-to-text), and mimo_tts (text-to-speech). Each tool uses a specific MiMo model; details are in the README table.
Why does the plugin need a Python driver?
Multimodal payloads (like images and audio) are large base64 strings. The DSH bash shell caps captured stdout at 64KB, which would silently truncate large payloads. The Python driver handles all file reading, body assembly, and HTTP POSTing in a single Python process, using spec/response files on disk to avoid shell truncation issues.
Are there any limitations with the TTS tool?
Yes. The 'voicedesign' model (mimo-v2.5-tts-voicedesign) does not accept an 'audio.voice' field; instead it uses 'optimize_text_preview'. The TTS target text goes in the assistant message, while the voice description goes in the user message. Also, the web search tool costs per keyword round, so be mindful of usage.