AI Configuration Ecosystem Map
AI Configuration Ecosystem
All Claude Code configuration, documentation, and tooling in domus-captures is organized across four locations. Each serves a distinct purpose per the Aristotelian taxonomy.
| Location | Purpose | Content | Type |
|---|---|---|---|
Reference — how the system works, what exists |
CLAUDE.md inventory, patterns, context library, ecosystem map |
Codex (lookup) |
|
Experiential — what worked, with dates |
Claude Code patterns, prompting, hooks, skills, agent orchestration, rules |
Pattern Journal (dated) |
|
Implementation — the build, architecture, decisions |
Architecture, tools, customization, reference, workflows, proposals |
Project (iterative) |
|
Configuration — the actual files Claude reads |
13 CLAUDE.md files, 8 context files, 3 global baselines |
Examples (source of truth) |
Layer Architecture
Global Layer (dots-quantum → ~/.claude/)
├── CLAUDE.md ← Role, style, invariants (872 lines)
├── settings.json ← Permissions, hooks, env vars
├── rules/ ← File-type rules (auto-loaded by extension)
│ ├── asciidoc.md ← Loaded for *.adoc files
│ ├── lua.md ← Loaded for *.lua files
│ ├── bash.md ← Loaded for *.sh files
│ ├── python.md ← Loaded for *.py files
│ └── d2.md ← Loaded for *.d2 files
├── skills/ ← Slash commands (7 skills)
│ ├── deploy.md ← /deploy spoke repos
│ ├── worklog.md ← /worklog daily capture
│ ├── project.md ← /project STD-001 management
│ ├── capture-idea.md ← /capture-idea → scaffold project
│ ├── save-commands.md ← /save-commands to quick-commands
│ └── session.md ← /session collaboration log
├── agents/ ← Specialized agents (5 agents)
│ ├── doc-auditor.md
│ ├── adoc-linter.md
│ ├── build-fixer.md
│ ├── project-modularizer.md
│ └── reviewer.md ← Read-only doc reviewer (parity with OpenCode)
└── hooks/ ← Event automation
├── pre-edit-backup.sh
├── post-edit-shellcheck.sh
└── session-banner.sh
Project Layer (per repo → .claude/)
├── CLAUDE.md ← Repo-specific context
├── agents/ ← Project-specific agents (optional)
└── settings.json ← Project overrides (optional)
Standards Governing AI Configuration
| Standard | What It Governs |
|---|---|
Default document-only posture, explicit delegation required |
|
Command quality criteria that AI output must meet |
|
Authoring rules AI must follow when writing .adoc files |
|
Credential rules AI must follow — never surface secrets |
Related
-
CLAUDE.md Patterns — what belongs in CLAUDE.md vs elsewhere
-
dots-quantum: AI Tools — stow packages for Claude configs