Open-Advisor

Open-source /advisor slash command for any AI CLI — consult models or CLI tools without leaving your session.

Runs as a plugin inside Qwen Code. Pick your advisor from the advisors array in ~/.qwen/settings.json.

/advisor.select nvidia-deepseek
/advisor Should I use StateFlow or SharedFlow for this ViewModel?

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 MiniMax M2.7 (NVIDIA) says:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

✓ MiniMax M2.7 (NVIDIA) responded (1247 chars)

Preview: Use StateFlow when the UI always needs the latest value (e.g. screen state).
Use SharedFlow for events that should be consumed once (e.g. navigation, toasts)…

Full response saved to: ~/.qwen/advisor/advisor-last-response.md

Say "show advisor response" to load the full reply.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

---

How it works

• 3 advisor types, pure Python runner — no Node.js, no browser, no AI interpretation:
  • model — calls modelProviders from settings.json via curl subprocess (OpenAI-compatible API)
  • cli — runs a local CLI tool (claude, gemini, etc.) via child process
  • http — direct HTTP call to any OpenAI-compatible endpoint via curl subprocess
• Switch any time with /advisor.select
• Every response is saved to <host-dir>/advisor-last-response.md; only a 200-char preview is shown inline

---

Prerequisites

• Python 3.10+
• Qwen Code (the host CLI)
• One of the advisor services (NVIDIA API, OpenRouter, local CLI tools, etc.)

Verified on Windows 11; should work on macOS and Linux.

Optional CLI advisor tools

Only needed if you want to consult one of these as your advisor:

Advisor	Install command	Auth
Claude Code	npm i -g @anthropic-ai/claude-code	Anthropic subscription
Gemini CLI	npm i -g @google/gemini-cli	Google account

---

Installation

git clone https://github.com/mohitsoni48/open-advisor.git
cd open-advisor

node install.mjs --ai <host>

<host> is one of:

--ai	Install dir	Commands subdir	Context file
qwen	~/.qwen/advisor/	commands/	QWEN.md
all	every host above	—	—

For each selected host the installer:

• Renders and copies /advisor, /advisor.select, /advisor.setup into the host's commands subdir
• Copies advisor.py and advisor-active into the host dir
• Writes the host's context file from the template, or appends an ## Advisor section if one already exists

The runner derives its host dir from ~/.qwen/advisor/ at runtime — no path strings baked into source.

---

Configuration

Edit ~/.qwen/settings.json — the advisors array defines all available advisors:

{
  "advisors": [
    {
      "id": "nvidia-deepseek",
      "name": "MiniMax M2.7 (NVIDIA)",
      "type": "http",
      "baseUrl": "https://integrate.api.nvidia.com/v1",
      "envKey": "NVAPI_KEY",
      "model": "minimaxai/minimax-m2.7",
      "generationConfig": {
        "temperature": 0.7,
        "top_p": 0.95,
        "max_tokens": 16384
      }
    },
    {
      "id": "claude-code",
      "name": "Claude Code",
      "type": "cli",
      "bin": "claude",
      "args": ["-p", "{{QUESTION}}", "--dangerously-skip-permissions"]
    },
    {
      "id": "qwen3.6-35b-a3b",
      "name": "Qwen 3.6 35B MoE",
      "type": "model"
    }
  ],
  "env": {
    "NVAPI_KEY": "nvapi-<your-key>",
    "LM_KEY": "sk-lm-<your-key>"
  },
  "modelProviders": {
    "openai": [
      {
        "id": "qwen3.6-35b-a3b@?",
        "baseUrl": "http://localhost:1234/v1/",
        "envKey": "LM_KEY"
      }
    ]
  }
}

Advisor types

Type	How it works	Config fields
model	Calls modelProviders entry matching id via curl	id
http	Direct HTTP POST to any OpenAI-compatible endpoint	baseUrl, envKey, model, generationConfig
cli	Runs a local CLI tool as a child process	bin, args (supports {{QUESTION}} placeholder)

---

Usage

/advisor.select <name>
/advisor <your question>

Examples:

/advisor.select nvidia-deepseek

/advisor In my Android ViewModel I need to combine a Room Flow with a SharedFlow from a Service. Should I use combine() or collectLatest?

/advisor Is it better to use a single Activity with multiple Fragments or multiple Activities for this flow?

Type "show advisor response" afterward to load the full reply into the chat.

---

Files installed per host

Path (relative to host dir)	Purpose
commands/advisor.md	The /advisor slash command
commands/advisor.select.md	The /advisor.select slash command
commands/advisor.setup.md	The /advisor.setup slash command
advisor.py	Pure Python runner — dispatches model/cli/http
advisor-active	Plain-text file: name of the currently-selected advisor
advisor-last-response.md	Full text of the most recent response (rewritten each call)
advisor-context.md	Advisor guidance appended to the host's context file

---

Known limitations

• Python HTTP libraries hang on Windows — the runner uses curl subprocess instead of urllib/httpx/openai SDK
• CLI advisors: Run synchronously with a 180-second timeout. On Windows the runner uses shell: true for npm shims.
• HTTP advisors: Requires network + a valid API key. The runner does not stream.

---

License

MIT