Orchestrate multi-agent debates to generate and critique design proposals.
npx skills add https://github.com/jpeggdev/buildwithjpegg --skill debate-jpeggdev
Please help me install this Agent Skill. Skill: debate Source: https://github.com/jpeggdev/buildwithjpegg/tree/main/plugins/tribunal/skills/debate Command: npx skills add https://github.com/jpeggdev/buildwithjpegg --skill debate-jpeggdev
This Skill addresses the risk of single-perspective designs by facilitating a structured debate among multiple AI agents to explore, critique, and converge on the best solution before formal design review.
Initiate a debate for the current task using the 'debate' trigger.
High-intent search queries and answers about installing and using this skill.
A multi-agent debate validates software architecture by having independent agents propose solutions, critique each other's trade-offs, and synthesize a consensus. This structured approach ensures diverse technical perspectives and risk assessment before formal design review.
Multi-agent consensus building is a structured debate process where AI agents representing different roles critique design proposals to find the best solution. You need it during pre-design phases to assess risks and explore diverse technical perspectives before committing to an architecture.
To start a multi-agent design debate, initiate the process using the 'debate' trigger. Agents will independently generate design proposals, perform adversarial critiques, and facilitate convergence, with configuration managed via YAML and final approval required from the user.
Multi-agent design debate requires no external dependencies and operates independently. Configuration for the structured agent prompting, proposal generation, critique, and synthesis is managed entirely through YAML files.
The best way to get diverse technical perspectives is running a multi-agent debate where agents role-play as Architect, Security, or CTO to explore and critique solutions. This ensures all angles and risks are considered before formal review.
You should avoid using adversarial multi-agent debate for simple tasks that do not require diverse technical perspectives or complex risk assessment. It is designed for pre-design phases of complex features where single-perspective designs pose a significant risk.
mind-systemsExecute a six-phase reflex-loop workflow to iteratively refine artifacts.
guicauladaGenerate structured design documents with alternatives and trade-off analysis.
hilaryosborneFacilitate structured dialogue to resolve software architecture disagreements and reach consensus.
baekenoughOrchestrate adversarial reviews among three or more LLMs to identify design flaws.
