architecture-decision-record

Use when a feature, workflow, or behavior change is underspecified and multiple materially different solutions are plausible.

Use when reviewing code, configuration, dependencies, or data flows for vulnerabilities, secrets exposure, or broken trust boundaries.

Use when a defined change spans multiple files, layers, migrations, or independently verifiable delivery steps.

Use when creating, adapting, evaluating, or improving an Agent Skill and its trigger description.

Use when a bug, failing test, crash, performance regression, build failure, or unexplained behavior needs diagnosis.

Use when creating or restructuring tutorials, how-to guides, reference documentation, or conceptual explanations.

Use on Desktop when preparing local changes for a clean commit, branch handoff, release note, or pull request description.

Use on Desktop when safely classifying, renaming, moving, deduplicating, or organizing files inside an authorized workspace.

Use on Desktop when reviewing a diff, commit, branch, or local code for correctness, regressions, security, performance, and maintainability.

Use on Desktop when profiling, cleaning, comparing, or analyzing local CSV, TSV, JSON, or spreadsheet-like data with reproducible evidence.

Use on Desktop when mapping an unfamiliar local repository, explaining its architecture, or preparing evidence-backed onboarding notes.

Create decision-focused meeting minutes from notes or transcripts when the user needs a reliable record of outcomes and follow-up work.