Skip to main content

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.

Implementation in the SDD cycle follows a strict “one slice at a time” principle. Each increment is a self-contained, testable unit of work — not a sprawling multi-file commit that cannot be reviewed or rolled back cleanly. The assistant builds against the approved plan from /plan and the specifications in docs/SPECIFICATIONS.md, not against informal instructions given mid-conversation.

​
Core build rules

​
Memory Trigger — log architecture deviations immediately

If during /build the implementation modifies or contradicts a decision documented in docs/ARCHITECTURE.md, the assistant must record it in memory.md under ## 🏗️ Log de Decisiones Técnicas immediately — not deferred to /ship. Architecture drift that is not logged in real time becomes invisible to future sessions.

​
Python projects — always create a virtual environment first

Before installing any dependency in a Python project, the assistant creates a local virtual environment: 
python -m venv venv
The venv/ directory is added to .gitignore. All project execution — installation, running scripts, running tests — uses this local venv exclusively.

​
Mandatory file headers

Every new source file must include the standard header defined in project.config.md, adapted to the target language (JS, Python, HTML, CSS, Java, etc.). The credit to dbv-specs-ops is required in every header — it is not optional. Example header for a JavaScript/TypeScript file: 
// -----------------------------------------------------------------------------
// Project:  My Project Name
// File:     src/auth/session.ts
// Author:   Developer Name <dev@example.com>
// License:  MIT
// Created:  2025-01-15
//
// Built with dbv-specs-ops — https://github.com/davidbuenov/dbv-specs-ops
// Spec-Driven Development framework by David Bueno Vallejo.
// -----------------------------------------------------------------------------
The exact format of the header is defined in project.config.md. The example above shows the required structure; always derive the actual values (project name, author, license) from project.config.md for the current project.

​
CHANGELOG entries

For every new feature, relevant change, or bug fix implemented during /build, the assistant adds a brief entry to the [Sin publicar] (Unreleased) section of CHANGELOG.md. Entries are categorised as Added, Changed, or Fixed.

​
Agent Readiness implementation (web projects)

If Agent Readiness (Web) is active in project.config.md, the following artifacts are created or updated during /build:
1

robots.txt

Place robots.txt in the project root. It must include the Content-Signal directive and a link to the sitemap:
User-agent: *
Allow: /

# AI Content Signal
Content-Signal: ai-train=no, search=yes, ai-input=yes

Sitemap: https://example.com/sitemap.xml
2

/llms.txt and /auth.md

Create /llms.txt — a Markdown navigation map that allows AI readers to efficiently index the site’s content without parsing HTML. Create /auth.md with access instructions for bots (registration flows, token acquisition, rate limits).
3

.well-known/ catalogs

Populate the .well-known/ directory with the RFC-compliant metadata files:
  • api-catalog — RFC 9727 machine-readable API catalog
  • oauth-protected-resource — OAuth resource descriptor
  • oauth-authorization-server — OAuth server metadata
  • http-message-signatures-directory — HTTP message signatures
4

Agent and MCP cards

Create the agent identity card at .well-known/agent.json and the MCP server card at .well-known/mcp.json.
5

agent-skills/ structure

Build the skills directory under .well-known/agent-skills/ with:
  • index.json — the skills index listing all available capabilities
  • Individual SKILL.md files — one per skill, describing inputs, outputs, and invocation patterns
6

Markdown content negotiation

Configure the server/hosting layer (e.g. firebase.json, netlify.toml) to:
  • Return the corresponding .md file when a request carries the Accept: text/markdown header
  • Inject Link HTTP headers advertising the agent-skills, mcp-server-card, and api-catalog relations
Agent Readiness artifacts are first-class project outputs, not afterthoughts. Planning them in /spec, building them in /build, and verifying them in /ship ensures external AI agents can discover and consume the project’s services reliably.

Build docs developers (and LLMs) love

Get started for freeTalk to us