Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/LIDR-academy/lidr-specboot/llms.txt

Use this file to discover all available pages before exploring further.

Specboot ships three pre-built AI agent role definitions, each packaged as a Markdown file with a YAML frontmatter header followed by a detailed system prompt. These files live in ai-specs/agents/ and are the canonical source of truth for agent configuration. When you add Specboot to a project, Claude and Cursor automatically discover them through symlinks — .claude/agents/ and .cursor/agents/ both point back to ai-specs/agents/ — so every copilot finds the same definitions without duplication.

​
What Are AI Agents in Specboot?

An agent file is a single .md file with two parts:
  • YAML frontmatter — declares the agent’s name, description, model, color, tools, and invocation examples that help the orchestrating AI know when to delegate to this agent.
  • System prompt — the detailed instructions, expertise areas, development approach, and output rules that shape the agent’s behavior during a session.
ai-specs/agents/
├── backend-developer.md
├── frontend-developer.md
└── product-strategy-analyst.md
The .claude/agents/ and .cursor/agents/ directories are relative symlinks that point to ai-specs/agents/. This means you only ever edit files in ai-specs/agents/ — both IDE integrations pick up changes automatically. After renaming or moving an agent file, always verify that the symlinks remain intact.

​
How Agents Are Discovered and Invoked

When you open a project that contains Specboot, Claude Code and Cursor scan the agents/ folder they are linked to and register each agent automatically. The orchestrating AI reads each agent’s description field (and its inline Examples blocks) to decide when a sub-task should be delegated. For example, asking “Create a new interview scheduling feature following DDD” causes the orchestrating model to route the request to the backend-developer agent, because the description and examples in backend-developer.md match that intent.

​
When to Use Agents vs. Just Chatting

Talking directly to your AI assistant works well for quick questions, small edits, or exploratory conversations. Agents add value when a task:
  • Spans multiple architectural layers (e.g., domain entity → application service → Express controller)
  • Requires role-specific expertise that benefits from a focused, opinionated system prompt
  • Should produce a durable artifact (like an implementation plan saved to .claude/doc/)
  • Is complex enough that generic responses risk missing project-specific patterns

​
The Three Included Agents

Backend Developer

DDD TypeScript architecture expert. Proposes implementation plans across Domain, Application, Infrastructure, and Presentation layers. Saves plans to .claude/doc/{feature}/backend.md.

Frontend Developer

React component architecture expert. Plans service layers, hooks, React Router, and React Bootstrap patterns — never writes implementation code. Saves plans to .claude/doc/{feature}/frontend.md.

Product Strategy Analyst

Ideation and value proposition expert. Uses Sequential Thinking MCP to analyze product ideas, define personas, and develop Jobs-to-be-Done frameworks.

Skills Overview

Agents automatically load applicable skills from ai-specs/skills/ when a request matches a skill’s description. Learn how skills extend agent capabilities.

​
How Agents Relate to Skills

Agents and skills are complementary. An agent provides a persistent role identity — expertise, architecture principles, output format — while a skill provides a reusable workflow that can be invoked by any agent when it matches the request. According to base-standards.md Section 4, an agent loads applicable skills automatically based on the task at hand. For instance, the backend-developer agent might invoke the code-auditing skill during a review request, or the using-git-worktrees skill before starting a new feature branch.
Skills live in ai-specs/skills/ and are mirrored into .claude/skills/ and .cursor/skills/ via the same relative symlink pattern used for agents. See the Skills Overview for a full list of available skills.

​
Creating Custom Agents

To add a new agent role to your project:
  1. Copy an existing agent file as your starting point: 
    cp ai-specs/agents/backend-developer.md ai-specs/agents/devops-engineer.md
  2. Update the YAML frontmatter with your new agent’s identity: 
    ---
name: devops-engineer
description: Use this agent when you need to configure CI/CD pipelines, Docker containers, or infrastructure-as-code. Examples: ...
model: sonnet
color: orange
tools: Bash, Glob, Grep, LS, Read, Edit, Write, WebFetch, WebSearch
---
  3. Rewrite the system prompt with the expertise areas, development approach, output format, and rules specific to your new role.
  4. Verify symlinks — because .claude/agents/ and .cursor/agents/ point to ai-specs/agents/, your new file is immediately discoverable without any additional configuration.
The key frontmatter fields to update are:
FieldPurpose
nameMachine-readable identifier used when routing tasks
descriptionNatural language description + Examples blocks that teach the orchestrator when to delegate
modelsonnet for most tasks, opus for deep strategic reasoning
colorVisual label color in IDE agent UIs
toolsComma-separated list of tools the agent is allowed to call
For a deeper look at customizing agents and connecting them to your project’s docs/ context, see the Customization Guide.

Build docs developers (and LLMs) love

Get started for freeTalk to us