my-topology-engine
Топологический движок AI-агентов (MCP-сервер)
Сервер Model Context Protocol (MCP), предоставляющий детерминированный топологический движок рабочих процессов для AI-агентов. Вместо использования открытых, непредсказуемых промптов с персоной, этот сервер взаимодействует с Cloudflare Workers для выполнения 10-шаговых конвейеров смены состояний на периферии, возвращая клиенту структурированные снимки прогресса и шаблоны итераций.
Обзор
Этот MCP-сервер действует как мост между AI-клиентами (такими как Claude Desktop, Cursor или Glama) и вычислительным движком топологии на периферии, работающим на Cloudflare Workers.
Он позволяет AI-агенту отображать (аналогизировать) намерение человека в фиксированный 10-шаговый конвейер смены состояний. Он поставляет 10 выходных снимков одновременно вместе с адаптивным JSON-шаблоном для заполнения пропусков для следующей итерации.
Related MCP server: project-planner-mcp
Ключевые возможности
Соответствие протоколу: Реализовано с использованием официального Model Context Protocol (
@modelcontextprotocol/sdk).Детерминированная топология: Выполняет фиксированные шаги решения задач на Cloudflare Edge, устраняя галлюцинаторные циклы и снижая потребление GPU/токенов.
Человек в цикле (UX снимков): Возвращает 10 промежуточных снимков прогресса одновременно, позволяя пользователям просматривать временную шкалу и бесшовно выполнять откат.
Адаптивный шаблон промпта: Добавляет адаптивную JSON-схему в конец выходных данных для плавной совместной доработки промптов человеком и ИИ.
Предоставляемые MCP-инструменты
Этот сервер предоставляет подключенным AI-клиентам следующие MCP-инструменты:
1. run_topology_pipeline
Выполняет 10-шаговый детерминированный топологический конвейер на Cloudflare Workers и возвращает 10 снимков смены состояний вместе с JSON-шаблоном итерации.
Схема ввода (
inputSchema):task_description(строка, обязательно): Задача или намерение пользователя, которое необходимо обработать через топологию.filled_template(объект, необязательно): JSON-объект, содержащий параметры или значения для заполнения пропусков, предоставленные человеком или выведенные агентом.
Поведение и инструкции к промпту агента:
AI-агент сопоставляет высокоуровневый запрос пользователя с фиксированными шагами топологии движка.
Сервер вызывает API Cloudflare Workers для выполнения переходов состояний.
Возвращает структурированную JSON-нагрузку, содержащую 10 снимков и
appendix_template.AI-агент преобразует
appendix_templateв естественный разговорный формат, чтобы помочь человеку уточнить входные данные для последующих запусков.
Архитектура и схема взаимодействия
[Человек / AI-клиент (Claude, Cursor, Glama)] │ │ MCP-протокол (Stdio) ▼ [Этот MCP-сервер (контейнер Node.js)] │ │ HTTP POST (Edge REST API) ▼ [Движок Cloudflare Workers] └─ Выполняет топологию конечного автомата из 10 шагов и возвращает снимки
Переменные окружения
WORKER_URL: URL-адрес вашей конечной точки Cloudflare Worker (например,https://my-topology-engine.my-agent-api.workers.dev).
Начало работы
Локальный запуск через Docker
# Build the Docker image
docker build -t mcp-topology-server .
# Run the MCP container
docker run -i --rm -e WORKER_URL="[https://my-topology-engine.my-agent-api.workers.dev](https://my-topology-engine.my-agent-api.workers.dev)" mcp-topology-server
Installation in Claude Desktop / MCP Clients
Add the following configuration to your claude_desktop_config.json:
{
"mcpServers": {
"topology-engine": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"WORKER_URL=[https://your-worker.workers.dev](https://your-worker.workers.dev)",
"mcp-topology-server"
]
}
}
}Available Tools
1 toolexecute_topology_engineC
Executes a 10-step deterministic problem-solving topology engine to process tasks and generate state-transition snapshots.
| Name | Required | Description | Default |
|---|---|---|---|
| filled_template | No | Optional key-value parameters to constrain or guide the 10-step execution. | |
| task_description | Yes | The primary task or problem to be solved. |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations are absent, so the description carries the full burden. It mentions 'deterministic' and '10-step' but does not disclose what happens during execution, side effects, required permissions, or what the output 'state-transition snapshots' entail. The behavior is underspecified for a tool with no annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, concise sentence with no redundant information. It is front-loaded with the core action and key characteristics. However, it could benefit from a couple more sentences to clarify usage, but as is, it is efficient.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity of a 10-step engine and no output schema, the description is insufficient. It does not explain what 'state-transition snapshots' are, how the filled_template constrains execution, or what the tool returns. The tool's operation remains abstract.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100% for the two parameters, but the description adds minimal meaning beyond the schema. It names 'filled_template' and 'task_description' but does not explain how they influence the 10-step execution. Baseline 3 is appropriate as the schema covers parameter definitions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool executes a 10-step deterministic problem-solving topology engine to process tasks and generate state-transition snapshots. The purpose is specific and distinct, though it introduces jargon ('topology engine') that could be ambiguous without further context.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
There is no guidance on when to use this tool versus alternatives, nor exclusions or prerequisites. The description implies it should be used for tasks requiring a structured problem-solving approach, but it does not explicitly define appropriate use cases or situations to avoid.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
1 tool update
v1.0.0- First observed
execute_topology_engine
TDQS
Scored across 1 tool
With only one tool, there is no possibility of confusion between different tools. The tool's purpose is clearly described as executing a topology engine, leaving no ambiguity.
The single tool name follows a clear verb_noun pattern ('execute_topology_engine'), which is consistent and predictable. Even with one tool, the naming is well-structured and aligns with common conventions.
One tool is on the borderline of being insufficient for a typical server, as most servers require at least a few tools to cover related operations. However, if the server's sole purpose is to run this engine, the count could be justified, but it feels thin without additional context.
The server exposes only a single operation, lacking any configuration, query, or retrieval tools that would complement the execution workflow. This creates significant gaps for agents that may need to set up inputs or inspect outputs, making the surface incomplete for a broader task lifecycle.
Maintenance
Related MCP Connectors
MCP server for AI agents to plan, verify, and deploy Cloudflare-native apps.
Cloudflare Workers MCP server: ai-agent-scratchpad
Cloudflare Workers MCP server: agent-workflow-engine
Cloudflare Workers MCP server: ai-eval
Related MCP Servers
- FlicenseNot gradedqualityNot gradedmaintenanceA production-ready MCP server for AI agents, providing deep web research and RAG capabilities via Cloudflare Workers.-
- FlicenseNot gradedqualityDmaintenanceA remote MCP server deployed on Cloudflare Workers for hosting and exposing MCP tools to AI clients via a remote endpoint.-
- AlicenseNot gradedqualityAmaintenanceAn MCP server that orchestrates image generation across multiple pipelines (OpenAI, Recraft, Midjourney) with durable memory for project context, running on Cloudflare Workers.1Apache 2.0
- AlicenseAqualityBmaintenanceAn MCP server that exposes deterministic workflows as tools, allowing small models to reliably orchestrate APIs and other MCP servers with minimal parameters.211 npm2MIT