Skip to content

Architecture

github-actions[bot] edited this page Jun 17, 2026 · 6 revisions

Architecture

8-habit-ai-dev is a markdown-first plugin. The stable core is the skills/ directory; runtime-specific files only package or present those skills to Claude Code, Codex, and other markdown-capable agents.

Note

This page describes how the documentation and plugin surfaces fit together. It is not required reading for everyday use.

Core Boundary

The plugin owns workflow discipline:

  • skill routing and markdown guidance,
  • human checkpoints,
  • review and verification prompts,
  • release and operational documentation discipline,
  • validation scripts for the plugin content itself.

The plugin does not own runtime enforcement, irreversible-action authorization, cloud automation, compliance certification, or dynamic orchestration engines.

Loading Model

Layer Claude Code Codex
Entry point CLAUDE.md AGENTS.md
Skills skills/*/SKILL.md skills/*/SKILL.md
Resolver skills/RESOLVER.md skills/RESOLVER.md
Session hooks hooks/ hooks/hooks.json parsed at install (schema-pure, top level = hooks only); SessionStart may run via JSON adapter
Package manifest .claude-plugin/ .codex-plugin/ and .agents/plugins/marketplace.json

Skill Shape

Each skill is a markdown file with YAML frontmatter and a consistent body:

---
name: <skill-name>
description: >
  When to use this skill
user-invocable: true
prev-skill: <predecessor|any|none>
next-skill: <successor|any|none>
---

Workflow skills also define what they expect from the previous step and what they produce for the next step.

Workflow Graph

/research -> /requirements -> /design -> /breakdown
    -> /build-brief -> /review-ai -> /deploy-guide -> /monitor-setup

The graph is validated by repository tests so skill references do not silently drift.

Validation

Script Purpose
tests/validate-structure.sh Frontmatter, names, version sync, graph fields, packaging surfaces
tests/validate-content.sh Markdown integrity, internal links, ADR shape, doctrine checks
tests/test-skill-graph.sh Handoff graph integrity
tests/test-verbosity-hook.sh Claude hook output and token budget

Documentation Surfaces

Surface Role
README.md Repository overview
AGENTS.md Cross-agent operating protocol
CLAUDE.md Claude Code architecture reference
docs/wiki/ Source for GitHub Wiki pages
llms.txt Flat map for LLM indexing
docs/adr/ Architecture decision records

See Also

Clone this wiki locally