agentleFS
cedricgnanscedricgnans/forge-solo/skills/forge-decision-record/SKILL.md

forge-decision-record

Skill0 starsChanged 3 months agoMITView on GitHub

Records a structural technical decision as an ADR in MADR format — context, decision drivers, considered options with pros and cons, the chosen outcome, consequences and how to confirm it. Produces docs/decisions/NNNN-title.md. Use this skill whenever a decision is hard to undo, a technology or boundary choice is being settled, a design violates a constitution principle, or the user asks "why did we choose that" or wants a trade-off documented.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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:

    Write an ADR when at least one of these holds:

    line 17
  2. Distinctive:

    Do not write one for an interchangeable library, a naming convention, or anything you can undo in an afternoon.

    line 26
  3. Distinctive:

    Never edit an accepted ADR: Write a new one that supersedes it and update the old status with a link.

    line 94
  4. Distinctive:

    When an ADR owns a violation of a principle, it must name it, say why the exception is justified here, and state whether it is local or …

    line 101

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Architecture decision record (ADR)

An ADR answers a question you will ask in six months: why this choice, and what did we know at the time? It captures the reasoning, not just the conclusion — the conclusion is already in the code.

Working solo, the ADR replaces the team conversation that never happened. It is the only place where the rejected alternative stays visible, which stops you from re-exploring it every year.

When to write one

Write an ADR when at least one of these holds:

  • Undoing the decision would take more than a week of work.
  • It constrains future decisions (data schema, service boundary, authentication model, public exchange format).
  • It contradicts a constitution principle.
  • It was hard to settle, or the rejected option was seriously defensible.
  • It is counter-intuitive and someone — you — will want to "fix" it later.

Do not write one for an interchangeable library, a naming convention, or anything you can undo in an afternoon. A repository drowning in ADRs is no better documented than one with none.

Format

File docs/decisions/NNNN-imperative-verb-phrase.md, sequential numbering, lowercase with hyphens.

---
status: proposed | accepted | deprecated | superseded by ADR-NNNN
date: YYYY-MM-DD
---

# NNNN. <Title: the decision, in the imperative>

## Context and problem statement
<2-4 sentences, or a short story. Phrase it as a question if you can. Make the scope explicit:
which components are affected.>

## Decision drivers
- <a desired quality, a constraint, a force at play>
- <…>

## Considered options
- <Option 1>
- <Option 2>
- <Option 3>

## Decision outcome
Chosen option: "<option>", because <justification: only one meeting a knock-out criterion |
best across the drivers | …>.

### Consequences
- Good, because <quality improved>
- Good, because <…>
- Bad, because <effort, risk, debt accepted>
- Bad, because <…>

### Confirmation
<How you will know the decision was actually implemented: a test, a review, a command, a
metric. Deciding is not enough.>

## Pros and cons of the options

### <Option 1>
<One-sentence description.>
- Good, because <…>
- Neutral, because <…>
- Bad, because <…>

### <Option 2>
…

## More information
<Links, measurements, experiments, what you did not know when deciding.>

Writing rules

  • The title is the decision, not the topic. "Use PostgreSQL for application state", not "Database choice".
  • One decision per ADR. Two linked decisions make two ADRs that cite each other.
  • The downsides are mandatory. An ADR with no "Bad, because" is not a decision, it is a retrospective justification. If you can't find one, you haven't looked hard enough at the opposing option.
  • The rejected option must be argued honestly. A straw man makes the ADR useless: in two years you won't know whether the option was genuinely evaluated.
  • Never edit an accepted ADR. Write a new one that supersedes it and update the old status with a link. The history of decisions is as useful as the decisions.
  • Record what you didn't know. "At the time of this decision, real volume was unknown; the estimate came from framing." That sentence is what will let you revise without guilt.

Link to the constitution

When an ADR owns a violation of a principle, it must name it, say why the exception is justified here, and state whether it is local or whether the principle should be amended. Two exceptions to the same principle signal a badly written principle: propose rewriting it rather than working around it a third time.

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.