Allows querying Google Gemini models for second opinions by asking a single model or the whole panel
Allows querying OpenAI models for second opinions by asking a single model or the whole panel
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@mcp-second-opinion ask the panel to review my SQL query for performance issues"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Let your MCP-aware agent consult rival LLMs mid-conversation. One simple MCP server, four providers — OpenAI, Google Gemini, Anthropic, xAI/Grok — unified behind two tools.
pip install mcp-second-opinionor with uv:
uv add mcp-second-opinionRelated MCP server: AI Consultant MCP Server
Set at least one of these environment variables:
Variable | Provider |
| OpenAI |
| Google Gemini |
| Anthropic |
| xAI/Grok |
Providers without a key set are gracefully disabled — they appear in panel responses with a friendly error field rather than crashing the server.
Optional:
Variable | Default | Purpose |
|
|
|
| unset | Provider key to omit from panel ( |
|
| Per-provider timeout in seconds. |
|
| Cap on rival output length. |
Claude Desktop / Claude Code (mcp_settings.json or equivalent):
{
"mcpServers": {
"second-opinion": {
"command": "mcp-second-opinion",
"env": {
"OPENAI_API_KEY": "sk-...",
"GEMINI_API_KEY": "...",
"ANTHROPIC_API_KEY": "...",
"XAI_API_KEY": "..."
}
}
}
}ask_other_model(question, model, context?, system_prompt?)Ask one specific rival.
question: "is this regex correct? /^[a-z]+$/"
model: "gpt-5" | "gemini-2.5-pro" | "claude-opus-4-7" | "grok-4" | ...
context: optional — code snippets, prior reasoning, files
system_prompt: optional — persona/framingReturns:
{
"answer": "...",
"model": "gpt-5",
"latency_ms": 842,
"tokens": {"input": 47, "output": 92},
"cost_usd": 0.0014
}ask_the_panel(question, context?, system_prompt?)Fan out to every enabled provider in parallel.
question: "what's wrong with this approach?"
context: optional
system_prompt: optional — applied uniformly to all panelistsReturns:
{
"responses": {
"openai": {"answer": "...", "model": "gpt-5", "latency_ms": 800, "cost_usd": 0.001, "error": null},
"gemini": {"answer": "...", "model": "gemini-2.5-pro", "latency_ms": 750, "cost_usd": 0.0008, "error": null},
"anthropic": {"answer": null, "model": "claude-opus-4-7", "error": "skipped (self)"},
"grok": {"answer": null, "model": "grok-4", "error": "XAI_API_KEY not set"}
},
"total_cost_usd": 0.0018,
"total_latency_ms": 800
}The server is a thin layer over LiteLLM, which provides a unified OpenAI-shaped interface to all four providers. Cost calculation comes from LiteLLM's pricing table.
MIT
Ask a single rival LLM for a second opinion.
| Name | Required | Description | Default |
|---|---|---|---|
| model | Yes | Model name, e.g. "gpt-5", "gemini-2.5-pro", "claude-opus-4-7", "grok-4". Provider is inferred from the prefix. | |
| context | No | Optional code snippets, prior reasoning, or files to include. | |
| question | Yes | The question to ask the rival model. | |
| system_prompt | No | Optional persona/framing for the rival. |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations are absent, so the description must disclose behavioral traits. It only mentions 'ask' and 'second opinion' with no detail on side effects, authentication, response format, or any limitations. This is insufficient for an agent to predict the tool's behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
A single sentence that front-loads the core action and scope. It is extremely concise with zero redundancy, achieving maximum efficiency for its content.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with 4 parameters and no annotations, the description is too sparse. It lacks usage guidance, behavioral context, and any indication of what the output looks like (even though an output schema exists, the description doesn't reference it). An agent is left without enough information to confidently invoke it.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% with clear descriptions for all parameters (model, context, question, system_prompt). The tool description adds no extra meaning beyond the schema, so the baseline of 3 applies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb (Ask) and resource (single rival LLM) with the purpose of obtaining a second opinion. The word 'single' explicitly distinguishes it from the sibling tool 'ask_the_panel', making the intent unmistakable.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool versus the sibling 'ask_the_panel'. The description implies a single opinion but does not state conditions for choosing this over the panel, leaving the agent to infer from the tool name alone.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Convene a panel of all enabled rival LLMs in parallel.
Returns labeled answers from each enabled provider, with per-slot error fields for any that failed or are not configured.
| Name | Required | Description | Default |
|---|---|---|---|
| context | No | Optional code snippets, prior reasoning, or files to include. | |
| question | Yes | The question to ask all rivals. | |
| system_prompt | No | Optional persona/framing applied uniformly. |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full behavioral disclosure burden. It explicitly mentions parallel execution and per-slot error fields for failed or unconfigured providers, which is important operational context. It does not mention potential latency, cost, or rate limits, but for a straightforward aggregation tool these are not critical gaps.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences with zero fluff. The action and the return format are stated immediately, and the error-handling note is placed after the primary behavior. Every sentence earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with 3 parameters and an output schema, the description covers the essential behavior and error handling. It does not explain prerequisites like API configuration, but that is implied by 'enabled providers'. The return format is handled by the output schema, so the description is adequately complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so all three parameters are already well-documented in the schema. The description adds no new semantic value beyond what the schema provides, such as example formats or edge-case guidance. This meets the baseline for high coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb ('Convene') and resource ('a panel of all enabled rival LLMs'), and clearly differentiates from the sibling 'ask_other_model' by emphasizing the parallel, multi-provider nature. It also describes the return value as labeled answers with per-slot error fields, leaving no ambiguity about what the tool does.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage for gathering multiple perspectives from all enabled providers in parallel, which contrasts with the single-model sibling 'ask_other_model'. However, it does not explicitly state when to choose this over the alternative or provide exclusions, leaving some room for inference.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.1ask_other_modelask_the_panelScored across 2 tools
The two tools have clear boundaries: ask_other_model queries a single rival LLM, while ask_the_panel convenes all enabled LLMs in parallel. There is no meaningful overlap between the two modes.
Both tool names follow a consistent 'ask_<target>' pattern with distinct nouns. This makes the tool set predictable and easy to understand.
Two tools is on the thin side, but it exactly matches the server's stated purpose: get a single second opinion or a panel opinion. Each tool earns its place with no redundancy.
The core use cases of requesting a single rival model or all enabled models are fully covered. A minor gap is the absence of a way to select a specific subset of models, but that may be configuration-level rather than tool-level.
A second opinion for AI agents: one prompt across several live Gonka models + roles, one call.
Talk to your public-facing AI from any MCP client — Claude, ChatGPT, Cursor, Cline, Windsurf.
One MCP endpoint for Claude, GPT & Gemini: 100+ tools + no-code connectors + agent workers.
MCP delegation fallback for AI agents to discover capabilities, knowledge, tools, and collaborators.
