specly: AI Project Management Server
Project management server with workflow-based feedback control for LLM IDE tools, built with TypeScript, Drizzle ORM, and MCP.
TypeScript MCP Drizzle ORM AI Agents
Workflow Architecture
specly is a Model Context Protocol (MCP) server for agent-driven project management workflows.
I modeled project state as explicit workflow transitions that an LLM IDE tool can read and update through MCP.
Execution Model
The core design is state-machine based project management. Agents query the current state, perform the next allowed transition, and leave an auditable trace.
Engineering Focus
- Deterministic Workflows: Explicit state transitions for AI interaction.
- Concurrency Model: Stateless protocol design that avoids blocking the event loop.
- Auditability: Agent traces are logged to support governance-sensitive operational workflows.