engineering-practices

Purpose

Guide the agent to ask the right questions about engineering practices, based on the project context. The agent explores decisions with the user — never prescribes.

Basic practices (CI/CD, tests, code review) are an assumed baseline. This skill focuses on choices that vary by context: which platform, which strategy, which stack.

Baseline

These practices are expected in any project. Confirm existence before exploring decisions:

• CI/CD pipeline (build + tests + automated deploy)
• Code review via PR/MR
• Automated tests in CI
• Version-controlled code with branching strategy

If any baseline does not exist, flag as a gap:

I did not find mention of [practice]. This is expected as baseline.
Does it already exist or does it need to be created?

Do not suggest baseline as a decision. If everything is covered, confirm and move on.

Input

Before exploring decisions, collect the project's operational profile.

If envisioning exists

Extract:

• Team scale (small/medium/large)
• Domain/industry
• Identified technical pain points
• Non-negotiable constraints and principles

If envisioning does not exist

Ask:

To identify which engineering decisions are relevant, I need to understand:

1. How many developers will work on the project?
2. What domain/industry? (regulated, e-commerce, internal, etc.)
3. Do you already have a defined CI/CD platform? Which one?
4. Do you already have a defined branch strategy? Which one?

Context-Based Exploration Guide

Use this matrix to decide which topics to explore with questions. Do not present topics that have no justification in the context.

Large scale (50+ devs)

Questions to ask:

• "With this volume of developers, how do you coordinate merges today? Have you considered trunk-based with feature flags?"
• "Is there a CODEOWNERS model or module ownership? How does it work?"
• "Are the CI gates sufficient to prevent breakage in main with so many contributors?"

Topics at play: branch strategy, CODEOWNERS, CI gate rigor.

Medium scale (10-50 devs)

Questions to ask:

• "What branch strategy do you use? Have you experienced friction with it?"
• "How many approvals do you require on PRs? Does it work well or create a bottleneck?"

Topics at play: branch strategy, level of rigor in PR reviews.

Small scale (<10 devs)

Just confirm baseline. Do not force complexity where it does not solve any real problem.

Regulated domain (fintech, healthcare, government)

Questions to ask:

• "What compliance requirements affect the pipeline? Do you need SAST, DAST, dependency scan?"
• "Is there a requirement for environment segregation or audit trail?"
• "Is IaC mandatory by policy or is it the team's choice?"

Topics at play: security gates, IaC, environment segregation, compliance gates.

Observability pain mentioned

Questions to ask:

• "Do you already have some instrumentation or are you starting from scratch?"
• "What level of observability do you need? Would centralized logs suffice, or do you need distributed tracing?"
• "Have you already chosen an observability SDK/platform, or is that decision still open?"

Topics at play: observability stack, tracing, alerts.

DevOps/CI-CD pain mentioned

Questions to ask:

• "What CI/CD platform do you use? Is the pain with the platform itself or with the pipeline?"
• "How do you deploy today? Is it manual, partially automated, or fully automated?"
• "Have you had rollback issues? Is there a defined strategy?"

Topics at play: CI/CD platform, deploy strategy, rollback.

SRE/Operations pain mentioned

Questions to ask:

• "Do you use SLOs/SLIs today? Or is operations reactive?"
• "Is there an incident management process? Runbooks?"
• "What level of operational maturity is the team aiming for?"

Topics at play: SLOs/SLIs, error budgets, runbooks, incident management.

Cloud deploy

Questions to ask:

• "Will you use IaC? Do you already have a preference between Terraform, Bicep, Pulumi?"

…