dbv-specs-ops is a lightweight, open-source framework that transforms any AI assistant — Claude, Gemini, ChatGPT, Cursor, Windsurf — into a disciplined Senior Software Engineer. Instead of ad-hoc prompting, every development session follows a strict Spec-Driven Development (SDD) lifecycle with mandatory phase gates, adversarial review, security checks, and automated context management.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/davidbuenov/dbv-specs-ops/llms.txt
Use this file to discover all available pages before exploring further.
Quickstart
Set up dbv-specs-ops in your project in under 5 minutes.
The SDD Workflow
Learn the 6 mandatory phases: /spec → /plan → /build → /test → /code-simplify → /ship.
Platform Setup
Configure dbv-specs-ops for Claude Code, Gemini, Cursor, Windsurf, and more.
Core Concepts
Understand Spec-Driven Development, the Agent Harness, and context economics.
Why dbv-specs-ops?
AI coding assistants are fast — but without structure, they produce unmaintainable code, leak secrets, hallucinate dependencies, and lose context between sessions. dbv-specs-ops solves this by wrapping the AI in a Harness: a complete set of rules, templates, context files, and phase gates that enforce engineering discipline at every step.6-Phase Lifecycle
Every feature goes through Spec → Plan → Build → Test → Simplify → Ship. No shortcuts.
Persistent Context
memory.md and task.md keep the AI oriented across sessions, tabs, and token limits.Security by Default
Mandatory security review gates catch credential leaks, slopsquatting, and logic vulnerabilities.
Adversarial Review
Every plan is stress-tested by a built-in adversarial architect before a single line is written.
Multi-Platform
Works with Claude Code, Gemini CLI, Antigravity, Cursor, Windsurf, and ChatGPT Web.
Agent Readiness
Built-in support for
robots.txt, llms.txt, agent.json, MCP cards, and agent skills.How it works
Install the framework
Copy the dbv-specs-ops files into your project root. The activation file (
CLAUDE.md, GEMINI.md, etc.) is auto-loaded by your AI platform at session start.Configure your project
Fill in
project.config.md with your project name, author, license, and language stack. The AI reads this on every session to stay oriented.Run your first /spec
Tell your AI:
/spec [describe your feature]. It will write structured requirements to docs/SPECIFICATIONS.md before touching any code.What’s included
| File | Purpose |
|---|---|
CLAUDE.md / GEMINI.md / .windsurfrules | Platform activation files — auto-loaded by your AI |
docs/MASTER_PROMPT.md | The core rules, workflow, and enforcement layer |
project.config.md | Project identity, model routing strategy, file headers |
docs/SPECIFICATIONS.md | Requirements template for the /spec phase |
docs/ARCHITECTURE.md | Technical architecture + Agent Harness template |
docs/DESIGN.md | Visual design system template (for UI projects) |
memory.md | Persistent ADR log and qualitative context |
task.md | Operational task list + context snapshots |
docs/ADOPTION_PROMPT.md | Onboard SDD to an existing project |
docs/UPGRADE_PROMPT.md | Zero-collision upgrade agent for new versions |
dbv-specs-ops is free and open source under the MIT license. Created by David Bueno Vallejo.