Perspectives

io.github.backspacevenkatPublished by GitHub account backspacevenkatv1.8.70Updated Oct 24, 2026

Query GPT 5.2, Claude Opus 4.5, Gemini 3, Grok 4.1 simultaneously for AI perspectives

VerifiedSTDIODesktop onlyDeveloper ToolsAI & ML

Overview

AI-generated overview

Queries several frontier AI models at once through the Polydev service and returns their separate answers plus a consensus, from inside the IDE.

What you can do
  • Send one prompt to multiple AI models at once and get each model's answer back separately, with a consensus and recommendation.
  • Check the status of local CLI tools and re-run detection of installed CLIs.
  • Send prompts to local CLI tools with fallback between them.
When to use it
Useful when you want several model opinions on a debugging problem, design choice or code review without switching between chat apps, and are willing to route prompts through a paid third-party service.
Requirements
Runs locally via npx (Node.js) as a stdio process. Needs a Polydev account and an API token supplied in the POLYDEV_USER_TOKEN environment variable, plus network access to the Polydev service. Free tier is listed at 1,000 messages per month; a Pro tier is paid.
Before you install
Prompts, including any code you paste, are sent to the Polydev service and on to third-party model providers. The API token is a secret and should be kept out of shared config files. Usage is metered, and the paid tier costs money. The send_cli_prompt tool can drive local CLI tools on your machine.

Installation

In SourceWeft

  1. Open Perspectives in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Polydev - Multi-Model AI Perspectives

Get unstuck faster. Query GPT 5.2, Claude Opus 4.5, Gemini 3, and Grok 4.1 simultaneously — one API call, four expert opinions.

[npm version] [SWE-bench Verified] [License: MIT]

Why Polydev?

Stop copy-pasting between ChatGPT, Claude, and Gemini. Get all their perspectives in your IDE with one request.

MetricResult
SWE-bench Verified74.6% Resolve@2
Cost vs Claude Opus62% lower
Response time10-40 seconds

"Different models have different blind spots. Combining their perspectives eliminates yours."

Supported Models

ModelProviderStrengths
GPT 5.2OpenAIReasoning, code generation
Claude Opus 4.5AnthropicAnalysis, nuanced thinking
Gemini 3 ProGoogleMultimodal, large context
Grok 4.1xAIReal-time knowledge, directness

Quick Start

1. Get your free API token

polydev.ai/dashboard/mcp-tokens

TierMessages/MonthPrice
Free1,000$0
Pro10,000$19/mo

2. Install in your IDE

Claude Code
bash
claude mcp add polydev -- npx -y polydev-ai@latest

Then set your token:

bash
export POLYDEV_USER_TOKEN="pd_your_token_here"

Or add to ~/.claude.json:

json
{  "mcpServers": {    "polydev": {      "command": "npx",      "args": ["-y", "polydev-ai@latest"],      "env": {        "POLYDEV_USER_TOKEN": "pd_your_token_here"      }    }  }}
Cursor

Add to ~/.cursor/mcp.json:

json
{  "mcpServers": {    "polydev": {      "command": "npx",      "args": ["-y", "polydev-ai@latest"],      "env": {        "POLYDEV_USER_TOKEN": "pd_your_token_here"      }    }  }}
Windsurf

Add to your MCP configuration:

json
{  "mcpServers": {    "polydev": {      "command": "npx",      "args": ["-y", "polydev-ai@latest"],      "env": {        "POLYDEV_USER_TOKEN": "pd_your_token_here"      }    }  }}
Cline (VS Code)
  1. Open Cline settings (gear icon)
  2. Go to "MCP Servers" → "Configure"
  3. Add the same JSON config as above
OpenAI Codex CLI

Add to ~/.codex/config.toml:

toml
[mcp_servers.polydev]command = "npx"args = ["-y", "polydev-ai@latest"]
[mcp_servers.polydev.env]POLYDEV_USER_TOKEN = "pd_your_token_here"
[mcp_servers.polydev.timeouts]tool_timeout = 180session_timeout = 600

Usage

Natural Language

Just mention "polydev" or "perspectives" in your prompt:

"Use polydev to debug this infinite loop"
"Get perspectives on: Should I use Redis or PostgreSQL for caching?"
"Use polydev to review this API for security issues"

MCP Tool

Call the get_perspectives tool directly:

typescript
{  "tool": "get_perspectives",  "arguments": {    "prompt": "How should I optimize this database query?",    "user_token": "pd_your_token_here"  }}

Example Response

🤖 Multi-Model Analysis
┌─ GPT 5.2 ────────────────────────────────────────│ The N+1 query pattern is causing performance issues.│ Consider using eager loading or batch queries...└──────────────────────────────────────────────────
┌─ Claude Opus 4.5 ────────────────────────────────│ Looking at the execution plan, the table scan on│ `users` suggests a missing index on `email`...└──────────────────────────────────────────────────
┌─ Gemini 3 ───────────────────────────────────────│ The query could benefit from denormalization for│ this read-heavy access pattern...└──────────────────────────────────────────────────
┌─ Grok 4.1 ───────────────────────────────────────│ Just add an index. The real problem is you're│ querying in a loop - fix that first.└──────────────────────────────────────────────────
✅ Consensus: Add index on users.email, fix N+1 query💡 Recommendation: Use eager loading with proper indexing

Research

Our approach achieves 74.6% on SWE-bench Verified (Resolve@2), matching Claude Opus at 62% lower cost.

ApproachResolution RateCost/Instance
Claude Haiku (baseline)64.6%$0.18
+ Polydev consultation66.6%$0.24
Resolve@2 (best of both)74.6%$0.37
Claude Opus (reference)74.4%$0.97

Read the full paper →

Available Tools

ToolDescription
get_perspectivesQuery multiple AI models simultaneously
get_cli_statusCheck status of local CLI tools
force_cli_detectionRe-detect installed CLI tools
send_cli_promptSend prompts to local CLIs with fallback

Links

IDE Guides

License

MIT License - see LICENSE for details.

Built by Polydev AI
Multi-model consultation for better code

Source: README.md at commit 086d0a9

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v1.8.70LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts