ormus-solutions/librecopy-claude-code/adr-templates
- community
Checking that the SKILL.md file is still in the repository.
ADR reference: what an ADR is, the status lifecycle, numbering, the index file, when to write one, Nygard, MADR, and Y-statement formats, and how ADRs relate to C4 levels. Use when setting up or maintaining an ADR log.
Install
npx vskill@latest install ormus-solutions/librecopy-claude-code/adr-templatesSKILL.md
ADR Templates
Architecture Decision Record formats, status lifecycles, and architectural documentation patterns.
Knowledge Base
What is an ADR?
An Architecture Decision Record captures a single architectural decision along with its context and consequences. ADRs are:
- Immutable: Once accepted, an ADR is not modified. If a decision changes, a new ADR supersedes the old one.
- Sequential: Numbered in the order they are created.
- Lightweight: One page, one decision. Not a design document.
- Permanent: Even deprecated ADRs remain in the repository as historical records.
ADR Status Lifecycle
- Proposed: Under discussion. Not yet decided.
- Accepted: Decision made. Team is committed to this approach.
- Deprecated: No longer relevant (e.g., the feature was removed).
- Superseded: A newer ADR replaces this decision.
ADR Numbering
Convention: ADR-NNNN with zero-padded numbers.
ADR Index (README.md)
When to Write an ADR
Write an ADR when:
- Choosing a technology (language, framework, database, cloud service)
- Choosing an architectural pattern (monolith, microservices, event-driven)
- Establishing a convention (branching strategy, API versioning, error handling)
- Making a decision that is costly to reverse
- Making a decision that the team disagreed about (the ADR records the resolution)
Do NOT write an ADR for:
- Trivial choices ("use prettier for formatting")
- Temporary decisions ("use mock data until API is ready")
- Business decisions (those go in product docs)
ADR Format Variants
Nygard Format (minimal):
MADR Format (recommended, used by this plugin):
Y-Statement Format (for the decision line):
Example: "In the context of choosing a primary datastore, facing the need for ACID transactions and complex joins, we decided for PostgreSQL, to achieve data consistency and query flexibility, accepting that horizontal scaling will require read replicas."
Architecture Documentation Levels (C4 Model)
| Level | What it Shows | Audience | Update Frequency | |
|---|---|---|---|---|
| Context (L1) | System + external actors | Everyone | Rarely | |
| Container (L2) | Deployable units | Developers, DevOps | When adding services | |
| Component (L3) | Internal structure of one container | Developers | When refactoring | |
| Code (L4) | Classes and functions | Not usually documented | Auto-generated if needed |
Patterns
- One Decision Per ADR: Do not combine "Use PostgreSQL" and "Use Redis for caching" into one ADR. They are separate decisions with separate contexts.
- Link Superseding ADRs: When ADR-012 supersedes ADR-005, update ADR-005's status to "Superseded by ADR-012" and link it.
- Include Rejected Options: Future developers need to know what was considered AND rejected, so they do not re-propose it.
- Date Everything: ADR decisions are anchored in time. The constraints that drove a decision in 2023 may not apply in 2025.
- Keep It Short: If an ADR exceeds 2 pages, either split the decision or move detailed analysis to an appendix.
Anti-Patterns
- ADR Graveyard: Writing ADRs that nobody reads. Link to relevant ADRs from code comments and PR descriptions.
- Editing Accepted ADRs: Modifying accepted ADRs to reflect current reality. Create a new ADR instead.
- Retroactive ADRs Without Admitting It: Writing an ADR for a decision made 6 months ago as if it were current. Note "Recorded retroactively" in the status.
- No Options Considered: An ADR that only describes the chosen option. Without alternatives, it is not a decision record; it is a description.
- Vague Consequences: "This might have performance implications." Be specific: "Read queries on the orders table will require an index scan above 10M rows."
References
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/architecture-docs/skills/adr-templates/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ormus-solutions/librecopy-claude-code/adr-templates)