agentleFS
jocarrdjocarrd/crux-agents/skills/adr/SKILL.md

adr

Skill4 starsChanged 18 days agoMITView on GitHub

Record an ADR: a technical decision in the project's canonical format, in `decisions/index.md` if it crosses repos or in `repos/<repo>/decisions.md` if not, with the context that motivated it, its consequences and its reversal cost. Use when an architecture decision is made, a choice is made between alternatives, an option is discarded for a non-obvious reason, when about to edit a decision log or mark one as superseded, and when someone asks why something is done this way.

90 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use fdr, sweep and 9 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • thiagofgfSkill

    fdr

    Use when creating, changing, reviewing, retiring, auditing, or backfilling implemented user-visible feature behavior, permissions, feature-specific rationale, tradeoffs, or relationships to architectural decisions.

    Also in 3 projects that use this

  • drewdrewthisSkill

    sweep

    Search for similar patterns, anti-patterns, or issues elsewhere in the codebase after a fix or feature. Use after implementing a fix to catch the same bug in other places, or proactively to audit for a known pattern.

    Also in 4 projects that use this

  • vndeeSkill

    product-spec

    Use when starting a new feature, product, or project — before any architecture or code, to define what to build and why

    Also in 3 projects that use this

  • OpenAISkill

    threat-model

    Use when Codex is already in the threat-modeling phase of a security scan, the user explicitly invokes $threat-model, or the user explicitly asks to create, update, or persist a repository threat model. Do not use as the primary trigger for full PR, commit, branch, patch, or repository scans.

    Also in 6 projects that use this

  • montymiSkill

    tech-spec

    Parse and review Technical Specification documents for completeness, consistency, and implementation readiness

    Also in 3 projects that use this

  • PydanticSkill

    coverage

    Fetch coverage diff from Codecov for the current branch or a specific PR. Shows uncovered lines, patch coverage, and overall coverage change.

    Also in 4 projects that use this

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 75 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When in doubt, start with single-repo.

    line 38
  2. Very specific:

    When a decision stops being valid it is neither edited nor deleted: it is marked Status: Superseded by <ref> and a new one is written.

    line 79

“Common” is measured: kept by at least 6 in 10 of the 75 edited versions of this file we could read. “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

adr

A decision that only lives in the head of whoever made it gets argued again every six months, or worse: it gets reverted by accident.

When it deserves a record

Yes:

  • A choice was made between real alternatives and the chosen one is not the obvious one.
  • Something was discarded for a reason not deducible from the code.
  • It affects how things will be written from now on.
  • Someone is going to ask "and why didn't we do X?".
  • Debt was knowingly taken on.

No:

  • Implementation details the code already explains.
  • Style preferences, which go to the conventions.
  • Decisions reversible in five minutes and with no consequences.

The filter: if a year from now someone could undo it for not knowing why it was done, it deserves a record.

Where it goes

ScopeLog
Affects more than one repodecisions/index.md
Affects a single reporepos/<repo>/decisions.md

When in doubt, start with single-repo. Promoting it later is provided for in the format; splitting up a badly placed cross-repo one is not.

Numbering is local to each log. To cite across logs, use the path prefix: decisions/ADR-013, repos/<repo>/ADR-007.

The format

### ADR-NNN: One-line title

- **Date**: YYYY-MM-DD
- **Status**: Proposed | Accepted | Superseded
- **Context**: what problem or situation motivates it
- **Decision**: what we decided
- **Consequences**: how the code, operations or deployment change
- **Reversibility**: how costly it would be to undo
- **Related**: refs to other ADRs

How to write each field

Context is the field that gets done badly. It doesn't describe the solution: it describes the situation that made it necessary, with data if there is any. Whoever reads it a year from now doesn't remember the problem, and without it the decision looks arbitrary.

Include here what was tried and didn't work, and the alternatives that were discarded with their reason. It is the part that prevents repeating the work.

Decision in the past tense and affirmative: what is done now, not what is proposed.

Consequences honest in both directions. An ADR that only lists advantages is justifying a decision already made. What is lost is written down too.

Reversibility is the most underrated field: it says whether this is a two-way or a one-way door, and therefore how much thought it needed.

Supersede, don't rewrite

When a decision stops being valid it is neither edited nor deleted: it is marked Status: Superseded by <ref> and a new one is written.

The log is history, not current state. Its value lies precisely in being able to see that something was decided, tried and changed, and why.

To promote from single-repo to cross-repo: a new entry in the cross-repo log, the old one marked as superseded, and the old one stays where it is.

When finished

A commit of its own, with the number and the title:

docs(decisions): add ADR-NNN <title>

Anti-patterns

  • Writing it afterwards, "when there is time". The context evaporates in days.
  • One ADR per change. The log stops being read if it is full of noise.
  • Writing in the future or conditional tense. An ADR records what was decided, not what is being thought about. That is what Status: Proposed is for.
  • Editing an old one so it fits today's reality. That destroys the log's only value.

Related

  • /promote-to-docs: for the rest of the information that has to survive
  • /end-session: where it gets detected that a decision went unrecorded

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.