Architecture Decisions (ADRs)

SkillAI & models

Architectural Decision Records, agent consume/cite/conflict; humans approve. Use for architecture-shaped work and ADRs.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Architecture Decisions (ADRs) skill

What this skill tells your AI

The instructions your AI receives, as published by nerds-odd-e/doughnut in .agents/skills/architecture-decisions/SKILL.md and read by ahel’s review.

  • Human playbook (propose / discuss / approve): docs/adrs/README.md
  • Accepted ADRs (current recommendations): enumerate the index in docs/adrs/README.md and confirm each record's Status — read explicitly; docs/ is outside default indexing. Filenames normally use *-accepted.md, but do not exclude Accepted records without that suffix.
  • Agent skill (use / cite / conflict / maintain): dough-adr-awareness in .agents/skills/dough-adr-awareness/ (Codex/Cursor) and .claude/skills/dough-adr-awareness/ (Claude Code).
  • Architecture-shaped work: cross-cutting stack, persistence, API contracts, authentication, packaging/monorepo layout, and shared conventions across backend/frontend/cli/mcp/e2e; explicit ADR work; or changes that reverse or bypass an Accepted ADR.
  • On architecture-shaped work, load relevant Accepted ADRs and follow them. Do not silently conflict. Do not approve ADRs; humans own the advice process.
  • Deviations require an explicit human-owned exception (record it in a PR/commit message or note pointing at the ADR) or a human superseding ADR. These constraints also apply to planning and delivery workflows.

Signals

GitHub stars
49
Forks
72
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
architecture-decisions-nerds-odd-e
Source
github.com/nerds-odd-e/doughnut