# Full Course: Spec-Driven Development with Coding Agents

## Executive summary

Spec-Driven Development (SDD) is presented as the disciplined, professional workflow for building complex applications using coding agents, moving beyond 'vibe coding.' SDD mandates that the developer focuses on writing a detailed specification (the 'what' and 'why') before the agent implements the code (the 'how'). The process involves establishing a project Constitution, defining feature specs, and following a structured loop: Plan $ ightarrow$ Implement $ ightarrow$ Validate $ ightarrow$ Replanning. This approach minimizes context decay, improves intent fidelity, and keeps the human developer in the role of the senior architect and supervisor.

## Key takeaways

- Three Main Benefits of SDD: SDD allows control of large code changes with small spec changes (e.g., changing from SQLite to MongoDB). It eliminates context decay between sessions, preserving non-negotiable context. Finally, it improves intent fidelity by forcing the definition of problem, success criteria, and constraints.
- The SDD Workflow Loop: The process is iterative: 1) Define the Project Constitution (Mission, Tech Stack, Roadmap); 2) Plan the Feature; 3) Implement the Feature; 4) Validate the Result; 5) Replanning (updating the Constitution/Roadmap). This loop is crucial for managing versioning and reducing cognitive debt.
- Agent Standards and Automation: The workflow can be automated using 'Skills' (a package of instructions/resources) and benefits from industry standards like the Agent Client Protocol (ACP) and Model Context Protocol (MCP), allowing developers to switch between different coding agents (e.g., Claude Code, Codeex) while maintaining the SDD workflow.

## Technical details

- Project Constitution: A foundational document defining the immutable standards, including the Mission (the 'why'), the Tech Stack (e.g., TypeScript, React, Next.js), and the Roadmap. This guides all future feature development.
- Feature Specification (Feature Spec): A detailed document created for each feature, containing the Plan (approach/sequence), Requirements (technical constraints), and Validation (success criteria). This spec is the contract between the human and the agent.
- Agent Standards and Tools: The workflow is supported by standards like the Agent Client Protocol (ACP) and Model Context Protocol (MCP). Tools include IDEs like WebStorm, coding agents (e.g., Claude Code), and version control via Git. Skills automate repeatable steps like generating a change log.
- Brownfield/Legacy Projects: SDD can be applied to existing codebases (brownfield) by having the agent reverse-engineer artifacts (like to-do files) to generate the initial Constitution and Roadmap, aligning future changes with past development.

## Practical implications

- Developers must shift their focus from writing code to writing clear, detailed specifications, acting as the 'senior architect' or 'supervisor' for the coding agent.
- The process emphasizes frequent, small commits and structured versioning of both code and specs to manage 'cognitive debt' and prevent drift.
- The ability to automate the SDD workflow using 'Skills' allows for repeatable, standardized processes across different projects and teams.

## Topics

Spec-Driven Development (SDD), Agentic Workflow, Software Architecture, Code Review, AI Coding Agents, Software Development Life Cycle (SDLC), JetBrains, DeepLearningAI, WebStorm IDE

Source: https://www.youtube.com/watch?v=hy8UstR2NEg