engineering-decisions

Resolves implementation and design decisions using canonical project precedence rules.

1|Updated Sep 11, 2026
One-click install
npx skills add https://github.com/rebuildup/my-web-2026 --skill engineering-decisions-rebuildup
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: engineering-decisions
Source: https://github.com/rebuildup/my-web-2026/tree/main/skills/engineering-decisions
Command: npx skills add https://github.com/rebuildup/my-web-2026 --skill engineering-decisions-rebuildup

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It prevents agents from escalating trivial implementation choices to users or making ad-hoc decisions that contradict project architecture, by enforcing a canonical decision precedence and clear escalation criteria. ## Core Features & Use Cases - Decision Precedence: Ranks canonical sources from project-wide policy down to local judgment, so conflicting guidance resolves deterministically. - Escalation Gating: Defines exactly when user confirmation is required, such as irreversible operations, merge authorization, external API contracts, or security risk acceptance. - Lifecycle Rules: Covers ADR supersession tracking, design-first workflows, dependency adoption checklists, naming responsibility, and early-stage compatibility policy. - Use Case: When an agent finds two conflicting conventions in a codebase, it checks ADRs and architecture docs first, follows the newer canonical source, and only asks the user if product semantics genuinely change. ## Quick Start Use the engineering-decisions skill to decide how to implement this change without asking me unnecessary questions.

Frequently Asked Questions about engineering-decisions

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I decide between conflicting coding conventions in a project?

Check canonical sources in precedence order: project-wide policy and architecture first, then design specs, then the majority consistent pattern in existing implementation. When sources conflict at the same level, prefer the more specific and newer canonical document.

When should an AI agent escalate a decision to the user?

Escalate only for genuine product or architecture decisions: irreversible operations, external API contract changes, security or privacy risk acceptance, meaningful cost increases, release scope changes, and merge or landing authorization. Resolve everything else from project evidence.

Can an agent merge a pull request after CI passes?

No. Passing reviews, CI, and acceptance criteria does not grant merge authorization. Merging, squashing, rebasing, or enabling auto-merge requires explicit user authorization for the identified pull request.

How should ADRs be updated when a decision is superseded?

Link the new ADR to the previous one and update the previous ADR's status with a superseded-by reference, keeping the decision graph traceable in both directions. Never silently delete a policy section without recording the revision.

Should I keep backward compatibility layers in an early-stage project?

Only when repository or release evidence shows a real compatibility obligation, such as a public API, persisted user data, or an explicit migration requirement. Otherwise migrate directly to the canonical design without speculative shims.