deliberation-engine

Multi-model consensus works by collecting independent model responses, running anonymous peer review, and synthesizing a final answer. You can trigger this structured deliberation pipeline by asking your question and including the /deliberate command.

Comparing AI perspectives for architecture decisions is handled by running structured multi-model consensus. The pipeline collects independent responses, ranks peer reviews, and synthesizes a single reliable recommendation to resolve technical tradeoffs.

Yes, multi-model deliberation coordinates OpenRouter model dispatch directly. A smart-router evaluates whether full deliberation is needed or if a single best model is sufficient, helping control API costs and optimize routing.

Yes, you can inject operational context from an Obsidian vault to enrich prompts during multi-model peer review. The pipeline optionally integrates this context when available and logs outcomes for continuity.

Yes, Python3 is required as a dependency to execute the multi-model consensus pipeline. The deliberation engine relies on Python scripts like openrouter_dispatch.py to manage model dispatch and peer review stages.

You should avoid multi-model consensus for simple queries where a single best model is sufficient. The built-in smart-router evaluates prompt complexity to bypass full deliberation, saving time and reducing API costs when heavy peer review is unnecessary.