repo-decisions
Skill10 starsChanged 32 days agoMITView on GitHub
Use when a change chooses among meaningful alternatives whose rationale may be revisited, or when writing, updating, reviewing, or superseding a MADR record
In this repository6 other files it gives its agents.
- AGENTS.mdAGENTS.md
- docs/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- repo-governanceSkill
- repo-reviewSkill
- repo-seedSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
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
- Distinctive: line 18
Write the required sections in order:
- Very specific: line 33
Superseding: never rewrite an old record into its opposite.
≈ “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.
Files
Writing and updating decision records
The decision log in docs/decisions/ is the durable home for architecture and process choices with genuine alternatives. Its directory entry defines the local or external standard; this skill carries the procedure.
When a decision record is required
Write or update a decision record when a change chooses among meaningful alternatives and future maintainers may reasonably revisit the rationale. Do not create an ADR for routine implementation, a mechanical refactor, an obvious fix already defined by a regression test, or work that merely follows an accepted decision. Risk-boundary feature behavior belongs in a spec; commits retain change history.
Procedure: create a new record
- Choose the next unused number in
docs/decisions/, normallymax + 1, zero-padded to 4 digits. Account for known parallel reservations. IDs never change; gaps are allowed. - Create
docs/decisions/NNNN-kebab-case-title.md. - Write the required sections in order:
## Status— first line is the status value; optionalClass: <value>on the next line.## Context and Problem Statement## Decision Drivers## Considered Options## Decision Outcome## Pros and Cons of the Options## Links
- State the decision, what it beats, and what it gives up.
## Considered Optionslists genuine alternatives;## Pros and Cons of the Optionsrecords why the losers lost. When the record relies on external evidence or an implementation source, cite it descriptively with a stable URL, DOI, or versioned permalink in## Links; do not leave research or quantitative claims unlinked. - Cross-reference records with relative Markdown links (
[0001](0001-title.md)), never bare numbers. - Run
node scripts/verify-decisions.mjs.
Procedure: update an existing record
- Status change to Accepted: the decision shipped. Keep the record current with what actually shipped (facts only — names, paths, structure — not the decision itself). Present tense.
- Superseding: never rewrite an old record into its opposite. Create a new record, mark the old one
Superseded by [NNNN](NNNN-title.md), and cross-link both. - Deprecation: a record that is no longer recommended but has no single successor becomes
Deprecated. - Rejected: a proposal considered and declined. Keep it while its rationale prevents a tempting mistake; otherwise delete it.
Procedure: respond to a verify-decisions failure
The verifier names the violation. Fix the named file:
missing status value— add the status line.invalid status— use Proposed/Accepted/Rejected/Deprecated orSuperseded by NNNN.invalid Class— use architecture/process/testing/feature/bug-fix/simplification.required sections missing or out of order— add or reorder the seven required sections.superseded target missing— create the target record or point at an existing number.duplicate decision number— choose an unused ID for the unmerged record and update its links; never renumber a published record.
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
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.

