agentpluginsdirectory
slior avatar

dialectic

v0.2.0

by slior · author: Lior Schejter

Configuration-first multi-agent design debate. Agents with different roles (architect, security, performance, simplicity,...) run structured debate rounds and a judge decides when convergence is reached.

What dialectic does, in the publisher's words

A multi-agent design debate system implemented entirely as agent skill files, prompt templates, and a JSON config. No code required. Runs on any agentic platform that supports subagent dispatch (Cursor, Claude Code, etc.).

Give it a design problem. Multiple AI agents, each with a distinct expert role (architect, security, performance, simplicity), debate the problem through structured rounds of proposals, critiques, and refinements. A judge evaluates convergence after each round. The debate ends when the agents reach sufficient agreement or a round ceiling is hit. A judge synthesizes the final solution.

The entire system is driven by an agent following skill files. State lives in plain markdown and JSON files you can read at any time.

From the project README, punctuation lightly normalized · Open the README on GitHub

Skills
2
MCP servers
0
Stars
4
License
MIT
Repo created
2026-02-20
Last pushed
2026-08-15
Publisher type
User
Version
0.2.0

Links

Skills · 2

judge
Evaluates convergence across agent refinements and, when the debate ends, writes the final synthesized solution. Operates in two modes: convergence_check (per-round verdict) and synthesis (final output). Use when the orchestrator needs a convergence verdict or a final synthesis document.
orchestrate
Drives the full multi-agent design debate lifecycle: loads config, initializes the workspace, runs optional clarification, executes debate rounds (proposal, critique, refinement), dispatches the judge for convergence checks, and writes the final synthesis. Use when the user wants to start or run a…

Descriptions come from the frontmatter of each SKILL.md, punctuation lightly normalized.

Category

Agent Tooling. Assigned by this directory. Agent Plugins 1.0.0 has no category field, so no manifest declares one.

Keywords

multi-agent · design · debate · architecture · review · consensus · orchestration

Related by publisher and keywords

  • axiomphni3j9a

    Codex engineering with Astra XHIGH planning and advice, Luna MAX Fast workers, Sol MAX design, independent Sol review, and context isolation.

  • maestraphni3j9a

    Explicitly activated, context-isolated multi-agent execution with Main-owned planning and bounded review.

  • caoawslabs

    Drive CLI Agent Orchestrator multi-agent sessions from any Agent-Plugins-compatible client. Prerequisites: the `uv` toolchain must be on PATH (the MCP server is launched via `uvx`), and a CAO API server must be running locally at http://127.0.0.1:9889 (`cao-server`). All communication is localhost-only.

  • bernsteinsipyourdrink-ltd

    The open-source governance layer for AI agents. Spawn parallel agents in per-task git worktrees, verify their work, and merge results; no model in the coordination loop, so replaying a plan reproduces its task graph byte-identically.

  • cli-agent-bridgeMiniMax-AI

    Delegate coding tasks from MiniMax Code to locally installed coding CLIs (Claude Code, Codex, Kimi Code, ZCode, DSH) through a dependency-free stdio MCP server with git-diff review.

  • graphflowRoarpeng

    图谱上下文压缩与任务规划插件（10 个 MCP 工具）。能力：压缩上下文、DAG 规划、增量建图、技能飞轮、诊断与产物导入导出。DeepSeek Harness：dsh plugin --profile web add @roarpeng/graphflow；工具名为 mcp__graphflow__graphflow_*。先调 graphflow_context（传 rootDir），复杂任务再 plan，改完代码后 index。

Add the verified badge to your README[![Verified by AgentPluginsDirectory.com against Schema 1.0.0](https://agentpluginsdirectory.com/badge.svg)](https://agentpluginsdirectory.com/plugins/dialectic)

✓ Verified . We fetched the manifest from GitHub and checked it against the official Agent Plugins 1.0.0 schema at agent-plugins.org.