dialectic-partner

To uncover design blind spots, run a structured dialectic between a main agent and a sub-agent. This process systematically challenges implied assumptions and identifies boundary risks by selecting appropriate sub-agent types to explore alternative perspectives.

Assumption testing in multi-agent decision support involves a main agent and sub-agent exchanging iterative critiques. This structured debate mechanism systematically challenges implied assumptions to surface hidden reasoning blind spots and boundary risks before finalizing important decisions.

Yes, you can use structured debate for git diff review by providing file paths and optional git diff output as external context. The sub-agent then critiques the code changes constructively, exploring boundary cases and alternative implementation perspectives.

No, the multi-agent debate includes a safety guardrail that prevents the main agent from changing code or files based on sub-agent suggestions. Explicit user approval is required before any unapproved code changes are executed, ensuring safe decision-quality evaluation.

To provide external context for a multi-perspective design evaluation, supply relevant file paths and optionally run git diff for code-change review. This mandatory provisioning allows the sub-agent to critique constructively and select appropriate evaluation perspectives.

Yes, multi-agent decision support persists each iterative debate round into a dedicated, timestamped record. This ensures full traceability and follow-up capability for decision-quality evaluation, allowing users to review the structured dialectic process later.