Generate multiple design perspectives in parallel and select the best candidate.
npx skills add https://github.com/AbyssCN/xihe --skill council-abysscn
Please help me install this Agent Skill. Skill: council Source: https://github.com/AbyssCN/xihe/tree/main/skills/council Command: npx skills add https://github.com/AbyssCN/xihe --skill council-abysscn
The council Skill helps users explore a wide solution space for a design/decision by generating multiple perspectives and allowing for a judge-and-select process.
Run the council skill with the command "council my design decision".
High-intent search queries and answers about installing and using this skill.
To evaluate multiple design options, you can generate parallel perspectives using different personas and apply a multi-lens judge to compare candidates and select the best decision. This approach thoroughly explores the solution space.
The best way to compare perspectives on a design decision is using a multi-lens judge to evaluate parallel-generated candidates. This dispatches different persona and angle combinations to produce options and selects the best one based on various viewpoints.
Use multi-perspective parallel generation for decision-making when facing complex design decisions that require exploring a wide solution space. It helps generate multiple options and choose the best one by evaluating candidates through different persona angles.
To start generating design candidates with different personas, run the command "council my design decision". This dispatches N different persona and angle combinations in parallel to produce candidates for multi-lens evaluation.
Yes, multi-judge option comparison works effectively for complex design decisions. It generates multiple perspectives in parallel and uses a multi-lens judge to evaluate the candidates, ensuring the best option is selected from a wide solution space.
rajnishkhatriExpand problem statements into candidate directions validated against repository evidence.
ebal5Evaluate design candidates with a 3D coupling model and generate ADR-ready documentation.
coval-aiAnalyze human versus LLM judge annotation discrepancies to refine evaluation prompts.
ZpankzEvaluate Claude Code commands, skills, and agents with structured rubrics.
