Execute one debate phase for a role-specific subagent in a multi-agent design debate.
npx skills add https://github.com/slior/dialectic-agentic --skill role-agent
Please help me install this Agent Skill. Skill: role-agent Source: https://github.com/slior/dialectic-agentic/tree/main/.cursor/skills/role-agent Command: npx skills add https://github.com/slior/dialectic-agentic --skill role-agent
This skill enables a single role-specific participant in a multi-agent design debate to generate a focused action (proposal, critique, or refinement) exactly once per phase, ensuring consistent participation and traceability.
Invoke the role-agent to perform a single phase (proposal, critique, or refinement) for a specific agent in the current debate workspace.
High-intent search queries and answers about installing and using this skill.
To generate deterministic outputs for a multi-agent debate, a role-specific agent executes a single phase action—proposal, critique, or refinement—exactly once, consuming metadata like AGENT_ID and ROUND to produce a focused, traceable artifact.
A multi-agent design debate is a workflow where distinct role-specific agents contribute perspectives like proposal, critique, or refinement. You need this when a design problem requires traceable, dialectic refinement across multiple specialized roles over multiple rounds.
You orchestrate a single debate phase by invoking the agent with metadata such as AGENT_ID, PHASE, and ROUND. The agent loads the appropriate role prompt, performs its designated action once, and writes the phase artifact to the debate workspace.
Yes, a single-role debate agent ensures consistent participation by performing exactly one well-scoped action per phase. It applies a defined role perspective, such as architect or security, to align with problem constraints and maintain determinism across rounds.
Yes, the role-agent workflow supports traceable artifacts by writing each phase output directly to the debate workspace. This ensures that every proposal, critique, or refinement action is stored and available for subsequent phases and analysis.
sliorGenerate role-specific clarifying questions from problem.md and role prompts.
sliorCoordinate multi-agent design debate rounds and write final synthesis.
jcolanoCoordinate multi-agent software development across the 17 SPARC modes.
tcytanGenerate GSD-compliant Claude Code command, skill, and agent files with type-specific validation.
