Comments
>log in to commentI built Kirat to give Claude a council of independent AI opinions—ChatGPT and Gemini—so you get a vote and a recorded dissent instead of a single answer. It runs locally, uses your own API keys, and is a one‑time purchase with no server. I’d love your thoughts; try the kirat_vote tool and let me know what breaks 😊
- Claude Desktop extension and Claude Code plugin
- Adds ChatGPT and Gemini as a council
- Independent model positions with dissent kept
- Cross‑checked code review by a second model family
- One‑time paid, bring your own API keys
Kirat is a Claude Desktop/Code extension that adds ChatGPT and Gemini to create a multi-model council that votes on your queries.
- for
- Developers and teams using Claude who want multi-LLM decision support.
- pricing
- paid
Key features
- Multi-model council — Calls Claude, ChatGPT and Gemini independently and aggregates their positions.
- Vote engine — Combines three model outputs into a majority verdict with recorded dissent.
- Budget guard — Enforces per-provider cost caps and daily spend limits.
- Failure handling — Retries on timeouts and provides fallback paths for unavailable models.
- Waiting room queue — Queues long-running requests and returns a ticket for later retrieval.
- No-server operation — Runs locally using your own API keys; nothing is sent to Kirat’s servers.
Use cases
- Choosing product pricing models (e.g., add a free tier)
- Selecting tech stacks (e.g., WooCommerce vs Shopify)
- Prioritizing small-business budget allocations
- Deciding architectural changes (e.g., monolith vs microservice)
- Generating ranked ideas or headlines
Kirat – Multi-Model AI Council vs alternatives
 Kirat – Multi-Model AI Council |  CrewAI |  LangGraph |  Mastra |  Claude Code | |
|---|---|---|---|---|---|
| Best for | Multi-model voting for decisions | Framework for custom multi-agent apps | Stateful agent workflow building | TypeScript AI agent framework | Agentic coding in Claude |
| Pricing | Paid | Subscription | Free | Free | Subscription |
| DevHunt upvotes | 1 | 0 | 0 | 0 | 0 |
| Launched | Oct 2026 | — | — | — | — |
- Kirat – Multi-Model AI Council vs CrewAI: CrewAI is a framework for building multi-agent systems, while Kirat is a ready-to-use extension that runs locally and focuses on voting.
- Kirat – Multi-Model AI Council vs LangGraph: LangGraph provides stateful agent orchestration code, whereas Kirat offers a pre-built UI and vote engine for Claude, ChatGPT and Gemini.
- Kirat – Multi-Model AI Council vs Mastra: Mastra is a TypeScript framework for AI agents; Kirat is a desktop plugin that aggregates existing LLMs without coding.
- Kirat – Multi-Model AI Council vs Claude Code: Claude Code is Anthropic’s coding assistant; Kirat extends Claude with additional models for decision voting, not just coding assistance.
Kirat – Multi-Model AI Council FAQ
How does Kirat obtain responses from the models?+
Kirat sends your prompt to Claude, OpenAI’s ChatGPT and Google Gemini using the API keys you provide, then collects the replies locally.
Does Kirat store any of my data?+
No. All requests are processed on your machine; nothing is sent to Kirat’s servers.
What costs are incurred per vote?+
Each vote incurs the API cost of the called models; the UI shows the total cost (e.g., $0.015) after the vote.
Can I use Kirat without an internet connection?+
Kirat requires internet access to call the external LLM APIs, but the orchestration runs locally.
Is there a free tier or trial?+
Kirat is sold as a one-time purchase; no free tier or trial is mentioned.
Summarized by DevHunt from kirat.ai · Oct 20, 2026. Details may change; check the official site.
