ADR Creator

Overview

This public intake copy packages

packages/skills-catalog/skills/(creation)/create-adr

https://github.com/tech-leads-club/agent-skills

Use it when the operator needs the upstream workflow, support files, and repository context to stay intact while the public validator and private enhancer continue their normal downstream flow.

This intake keeps the copied upstream files intact and uses

metadata.json

ORIGIN.md

ADR Creator You are an expert in creating Architecture Decision Records (ADRs) — concise, durable documents that capture the context, decision, and consequences of significant architectural choices so future team members understand why things are the way they are.

Imported source sections that did not map cleanly to the public headings are still preserved below or in the support files. Notable imported sections: ADR vs RFC — Critical Distinction, Language Adaptation, ADR Format Selection, Document Templates, Decision Drivers, Considered Options.

When to Use This Skill

Use this section as the trigger filter. It should make the activation boundary explicit before the operator loads files, runs commands, or opens a pull request.

• User asks to "write an ADR", "create an ADR", "add an architecture decision record"
• User wants to "document why we chose X", "record this decision", "capture this architectural choice"
• A significant technical decision has been made (or is being finalized) and needs to be recorded
• The team wants to preserve the reasoning behind a choice for future engineers
• User asks "why did we choose X" and the answer should be written down permanently
• Decisions not yet made — use create-rfc to drive the decision process first

Operating Table

metadata.json

ORIGIN.md

README.md

README.md

## Related Skills

Workflow

This workflow is intentionally editorial and operational at the same time. It keeps the imported source useful to the operator while still satisfying the public intake standards that feed the downstream enhancer flow.

1. Decision title (noun phrase, not a question — e.g., "Use Redis for session storage")
2. Date of the decision (or today's date)
3. Status (Accepted / Proposed / Deprecated / Superseded)
4. Context — the forces, constraints, and situation that made this decision necessary
5. The decision itself — what was chosen and why
6. Consequences — what becomes easier, harder, or different as a result
7. Decision drivers — the key criteria or constraints

Imported Workflow Notes

Imported: Interactive Workflow

Step 1: Gather Context (if not provided)

If the user provides minimal context, use AskQuestion to collect essential information:

{
  "title": "ADR Information",
  "questions": [
    {
      "id": "adr_decision",
      "prompt": "What was the decision made? (e.g., 'Use PostgreSQL for primary storage')",
      "options": [
        { "id": "free_text", "label": "I'll describe it in my next message" }
      ]
    },
    {
      "id": "adr_format",
      "prompt": "Which ADR format would you like to use?",
      "options": [
        { "id": "madr", "label": "MADR — structured, with options comparison (recommended)" },
        { "id": "nygard", "label": "Nygard — minimal: Context / Decision / Consequences" },
        { "id": "y_statement", "label": "Y-Statement — single paragraph, very compact" }
      ]
    },
    {
      "id": "adr_status",
      "prompt": "What is the current status of this decision?",
      "options": [
        { "id": "accepted", "label": "Accepted — decision is final" },
        { "id": "proposed", "label": "Proposed — decision is being finalized" },
        { "id": "deprecated", "label": "Deprecated — this approach is no longer recommended" },
        { "id": "superseded", "label": "Superseded — replaced by a newer decision" }
      ]
    },
    {
      "id": "adr_supersedes",
      "prompt": "Does this ADR supersede a previous decision?",
      "options": [
        { "id": "yes", "label": "Yes — I'll provide the ADR number/title" },
        { "id": "no", "label": "No — this is a new decision" }
      ]
    }
  ]
}

Step 2: Validate Mandatory Fields

MANDATORY fields — ask if missing:

• Decision title (noun phrase, not a question — e.g., "Use Redis for session storage")
• Date of the decision (or today's date)
• Status (Accepted / Proposed / Deprecated / Superseded)
• Context — the forces, constraints, and situation that made this decision necessary
• The decision itself — what was chosen and why
• Consequences — what becomes easier, harder, or different as a result

RECOMMENDED fields:

• Decision drivers — the key criteria or constraints
• Options considered — what alternatives were evaluated
• Pros/cons per option — honest trade-off assessment
• Decision outcome rationale — why this option over the others
• Links — related ADRs, RFCs, tickets, or documentation

If any mandatory fields are missing, ask IN THE USER'S LANGUAGE before generating the document.

Step 3: Assign ADR Number

Scan the existing ADR directory for the next sequential number:

1. Check if an ADR directory exists (

   docs/adr/

   ,

   docs/decisions/

   ,

   .adr/

   , or

   adr/

   )
2. Find the highest existing number
3. Assign the next number (e.g., if ADR-007 exists, this becomes ADR-008)
4. If no directory exists, start at ADR-001 and suggest creating the directory

Step 4: Generate the ADR

Generate the ADR following the format selected in Step 1.

Step 5: Offer File Placement

After generating, ask where to save it:

ADR Created: "ADR-{NNN}: {Title}"

Suggested file path: docs/adr/{NNN}-{kebab-case-title}.md

Would you like me to:
1. Save it to docs/adr/ (recommended convention)
2. Save it to a different location
3. Just show the content (I'll place it manually)

Imported: ADR vs RFC — Critical Distinction

If the user says "I need to decide whether to do X" → use

create-rfc

Examples

Example 1: Ask for the upstream workflow directly

Use @create-adr to handle <task>. Start from the copied upstream workflow, load only the files that change the outcome, and keep provenance visible in the answer.

Explanation: This is the safest starting point when the operator needs the imported workflow, but not the entire repository.

Example 2: Ask for a provenance-grounded review

Review @create-adr against metadata.json and ORIGIN.md, then explain which copied upstream files you would load first and why.

Explanation: Use this before review or troubleshooting when you need a precise, auditable explanation of origin and file selection.

Example 3: Narrow the copied support files before execution

Use @create-adr for <task>. Load only the copied references, examples, or scripts that change the outcome, and name the files explicitly before proceeding.

Explanation: This keeps the skill aligned with progressive disclosure instead of loading the whole copied package by default.

Example 4: Build a reviewer packet

Review @create-adr using the copied upstream files plus provenance, then summarize any gaps before merge.

Explanation: This is useful when the PR is waiting for human review and you want a repeatable audit packet.

Best Practices

Treat the generated public skill as a reviewable packaging layer around the upstream repository. The goal is to keep provenance explicit and load only the copied source material that materially improves execution.

• Keep the imported skill grounded in the upstream repository; do not invent steps that the source material cannot support.
• Prefer the smallest useful set of support files so the workflow stays auditable and fast to review.
• Keep provenance, source commit, and imported file paths visible in notes and PR descriptions.
• Point directly at the copied upstream files that justify the workflow instead of relying on generic review boilerplate.
• Treat generated examples as scaffolding; adapt them to the concrete task before execution.
• Route to a stronger native skill when architecture, debugging, design, or security concerns become dominant.

Troubleshooting

Problem: The operator skipped the imported context and answered too generically

Symptoms: The result ignores the upstream workflow in

packages/skills-catalog/skills/(creation)/create-adr

metadata.json

ORIGIN.md

Problem: The imported workflow feels incomplete during review

Symptoms: Reviewers can see the generated

SKILL.md

Problem: The task drifted into a different specialization

Symptoms: The imported skill starts in the right place, but the work turns into debugging, architecture, design, security, or release orchestration that a native skill handles better. Solution: Use the related skills section to hand off deliberately. Keep the imported provenance visible so the next skill inherits the right context instead of starting blind.

Imported Troubleshooting Notes

Imported: Context and Problem Statement

{Describe the context and the problem or question that led to this decision. 2–4 sentences. What situation forced this choice?}

Related Skills

• @accessibility

  - Use when the work is better handled by that native specialization after this imported skill establishes context.

• @ai-cold-outreach

  - Use when the work is better handled by that native specialization after this imported skill establishes context.

• @ai-pricing

  - Use when the work is better handled by that native specialization after this imported skill establishes context.

• @ai-sdr

  - Use when the work is better handled by that native specialization after this imported skill establishes context.

Additional Resources

Use this support matrix and the linked files below as the operator packet for this imported skill. They should reflect real copied source material, not generic scaffolding.

references

references/n/a

examples

examples/n/a

scripts

scripts/n/a

agents

agents/n/a

assets

assets/n/a

• README.md

Imported Reference Notes

Imported: Language Adaptation

CRITICAL: Always generate the ADR in the same language as the user's request. Detect the language automatically.

• Keep technical terms in English when appropriate (e.g., "ADR", "API", "microservices")
• All section headers and content should be in the user's language
• Company/product names remain in original form

Imported: ADR Format Selection

Three formats are widely used. Detect the right one from context, or ask:

Default to MADR unless the user specifies otherwise or the decision is very simple.

Imported: Document Templates

MADR Format (Default)

# ADR-{NNN}: {Title}

- **Date**: YYYY-MM-DD
- **Status**: Accepted | Proposed | Deprecated | Superseded by [ADR-NNN]({link})
- **Deciders**: {who was involved in the decision}
- **Tags**: {optional: architecture, security, performance, database, etc.}

#### Imported: Decision Drivers

- {Driver 1 — e.g., "Must support 10k concurrent users"}
- {Driver 2 — e.g., "Team has no Go experience"}
- {Driver 3 — e.g., "Must be deployable on-premise"}

#### Imported: Considered Options

- {Option A}
- {Option B}
- {Option C — "Do nothing / status quo" when relevant}

#### Imported: Decision Outcome

Chosen option: **"{Option A}"**, because {concise rationale tied to decision drivers}.

### Positive Consequences

- {Benefit 1}
- {Benefit 2}

### Negative Consequences

- {Trade-off 1 — be honest}
- {Trade-off 2}

#### Imported: Pros and Cons of the Options

### {Option A} ✅ Chosen

- ✅ {Pro 1}
- ✅ {Pro 2}
- ❌ {Con 1}

### {Option B}

- ✅ {Pro 1}
- ❌ {Con 1}
- ❌ {Con 2}

### {Option C}

- ✅ {Pro 1}
- ❌ {Con 1}

#### Imported: Links

- {Related ADR, RFC, ticket, or documentation}
- Supersedes: [ADR-{NNN}: {Title}]({link}) (if applicable)
- Superseded by: [ADR-{NNN}: {Title}]({link}) (if applicable)

Nygard Format (Minimal)

# ADR-{NNN}: {Title}

#### Imported: Status

Accepted | Proposed | Deprecated | Superseded by ADR-{NNN}

#### Imported: Context

{What is the situation that led to this decision?
What forces are at play — technical, business, organizational?
What constraints exist? 2–5 sentences.}

#### Imported: Decision

{What did we decide to do?
State it directly, in active voice: "We will use X" or "We decided to adopt Y."
Include a brief rationale — why this option over the alternatives.}

#### Imported: Consequences

{What becomes easier or better as a result?}
{What becomes harder or worse? Be honest about trade-offs.}
{What new concerns or constraints does this introduce?}

Y-Statement Format (Compact)

# ADR-{NNN}: {Title}

**Date**: YYYY-MM-DD | **Status**: Accepted

In the context of **{situation/use case}**,
facing **{concern or constraint}**,
we decided **{the option chosen}**,
to achieve **{quality attribute or goal}**,
accepting **{the downside or trade-off}**.

**Deciders**: {names or roles}
**Links**: {related ADRs, tickets}

Imported: ADR Quality Checklist

Before finalizing, verify:

• Title is a noun phrase describing the decision (not a question, not a vague label)
• Date is included (decisions without dates lose context quickly)
• Status is set correctly — Accepted, Proposed, Deprecated, or Superseded
• Context explains the forces that made this decision necessary, not just what was done
• Decision is stated directly and tied to the context
• Consequences include honest trade-offs — not just positives
• Options (MADR format) include at least 2 alternatives actually considered
• Supersedes / superseded by links are included when applicable
• File follows naming convention:

  NNN-kebab-case-title.md

• Number is sequential in the ADR directory

Imported: ADR File Naming Convention

docs/adr/
├── 001-use-postgresql-for-primary-storage.md
├── 002-adopt-event-driven-architecture.md
├── 003-replace-jenkins-with-github-actions.md   ← supersedes ADR-001 if relevant
└── README.md                                     ← optional index

• Zero-padded numbers:

  001

  ,

  002

  , ...

  099

  ,

  100

• Kebab-case title

• .md

  extension
• Common directories:

  docs/adr/

  ,

  docs/decisions/

  ,

  adr/

  ,

  .adr/

Imported: Common Anti-Patterns to Avoid

Title as a Question

BAD:

# ADR-001: Should we use PostgreSQL?

GOOD:

# ADR-001: Use PostgreSQL for Primary Storage

Titles should record the decision, not the question. Future readers need to know what was decided, not what was considered.

Vague Context

BAD:

We needed a database and chose PostgreSQL.

GOOD:

Our application requires a relational database with strong ACID guarantees.
The team has deep PostgreSQL experience. MySQL was evaluated but lacks
native support for JSONB columns, which our schema design requires.
Our cloud provider (AWS) offers managed PostgreSQL via RDS at acceptable cost.

Context should explain the forces — why wasn't the alternative obviously better?

Consequences Without Trade-offs

BAD:

#### Imported: Consequences

PostgreSQL is fast and reliable.

GOOD:

#### Imported: Consequences

- Enables JSONB columns and advanced indexing for our query patterns
- Team expertise means fast onboarding and fewer operational surprises
- Adds operational burden compared to a managed NoSQL service
- Schema migrations require careful planning in a relational model

Honest trade-offs are what make ADRs valuable years later.

Editing Instead of Superseding

BAD: Editing an old ADR to change the decision after the fact.

GOOD: Creating a new ADR with

Status: Superseded by ADR-{NNN}

ADRs are historical records. The old decision was correct given what was known at the time. Superseding preserves that context.

Missing the "Why Not" Rationale

BAD:

#### Imported: Decision

We will use Redis for session storage.

GOOD:

#### Imported: Decision

We will use Redis for session storage. We considered storing sessions in PostgreSQL
(already in our stack) but Redis's built-in TTL support and in-memory performance
make it significantly better suited for high-frequency session reads. The operational
cost of an additional service is justified by the simplified session expiry logic.

The rationale is why this option and not the others — not just what was chosen.

Imported: Important Notes

• ADRs are immutable — never edit the decision. Supersede with a new ADR.
• Short is better — 200–500 words is ideal. If it needs to be longer, move detail to a linked TDD or RFC.
• Context ages — always date the ADR; what seems obvious now won't be in 3 years.
• Honest consequences — a one-sided ADR loses credibility. Future engineers will hit the downsides regardless.
• Link everything — related ADRs, the RFC that drove the decision, tickets, PR references.
• Language adaptation — always write in the user's language.
• Number sequentially — check the directory before assigning a number.