agentleFS
Sign inStart free
jhonatan-oliveiradevjhonatan-oliveiradev/agent-skills/skills/documenting-architecture-decisions/SKILL.md

documenting-architecture-decisions

Skill1 starsChanged 3 months agoMITView on GitHub

Use when a software decision materially affects structure, quality attributes, dependencies, interfaces, technology choices, or team direction and its rationale must remain reviewable over time.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not create ADRs for routine implementation details that are cheap to reverse and already obvious from the code.

    line 20
  2. Distinctive:

    Keep each ADR focused on one decision and include:

    line 23
  3. Distinctive:

    If the record only names a technology and says it is “best”, the decision is not sufficiently documented.

    line 46

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Documenting Architecture Decisions

Core principle

Record why an architecturally significant choice was made while the context is still available. An ADR is a decision record, not a design essay and not retrospective justification.

When a decision deserves an ADR

Capture a decision when changing it later would be costly or when it materially affects:

  • system structure or deployment shape;
  • reliability, security, performance, availability, or other quality attributes;
  • dependency and ownership boundaries;
  • published interfaces or data contracts;
  • frameworks, persistence, messaging, infrastructure, or other construction techniques;
  • a recurring technical debate that needs one shared source of rationale.

Do not create ADRs for routine implementation details that are cheap to reverse and already obvious from the code.

Minimal record

Keep each ADR focused on one decision and include:

  1. Title and status — proposed, accepted, rejected, or superseded.
  2. Context — the problem, constraints, forces, and evidence that make a decision necessary.
  3. Decision drivers — the quality attributes or business/engineering outcomes that matter most.
  4. Considered options — realistic alternatives, including keeping the current state when relevant.
  5. Decision — the selected option in direct language.
  6. Consequences — benefits, costs, new risks, operational burden, and follow-up work.
  7. Validation/review trigger — evidence that should confirm the decision and conditions that should cause it to be reconsidered.

Lifecycle

Discuss the ADR while it is proposed. Once accepted, preserve it as historical context. If later evidence changes the decision, create a new ADR that supersedes the old one rather than rewriting history.

Link implementation changes to the relevant ADR when that connection helps reviewers evaluate architectural consistency.

Quality checks

A useful ADR lets a future engineer answer:

  • What problem were we solving?
  • Which constraints and trade-offs mattered?
  • What alternatives were actually considered?
  • Why did this option win then?
  • What complexity did we knowingly accept?
  • What would make us revisit the choice?

If the record only names a technology and says it is “best”, the decision is not sufficiently documented.

Reference

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free