liveBook · Manning

Review process

As explained in Automating API Delivery

An API review process is a defined workflow for evaluating and approving API designs against stated principles, standards, and business needs.

An API design can pass automated checks and still be difficult for consumers to use. A review process gives colleagues and potential users a way to assess the proposed operations, resource model, documentation, and business purpose before the design is implemented. Automation handles repeatable standards checks; people concentrate on decisions it cannot reliably judge. section 5.0 section 5.1 section 5.5.2

Start a design-stage review when an API definition change is submitted as a pull request (PR), after linting and breaking-change checks have run. Reviewers examine the change and its check reports, leave comments, and approve the PR when concerns are resolved. A small change that closely follows the style guide may need only asynchronous comments. A broad or complex change may need a meeting to establish context and resolve design decisions together. section 5.1 section 5.2.5

For a review meeting, use a repeatable sequence:

  • Invite the relevant participants, present an agenda, and establish the business context and the job API consumers need to perform. section 5.2 section 5.2.5
  • Examine the resource model and rendered API documentation. Check whether the design has the required operations and resources, whether any operations are missing, and whether the API boundary makes sense. section 5.2 section 5.2.5
  • Inspect automated reports for warnings needing attention, then discuss design concerns not covered by those checks and capture feedback on the PR. Note style-guide gaps or recurring problems that could become automated checks. section 5.1 section 5.2 section 5.2.5
  • Close by asking participants for a short survey; use the responses to assess the meeting rather than assuming that approval alone means the review was effective. section 5.2 section 5.7

Document who requests and facilitates reviews, how work is prioritized, and which changes require a meeting. Keep the process easy to find alongside the guidance teams use. Set goals for quality, speed, dependable scheduling, flexibility across small and large changes, and cost. These are choices to make for the organization, not universal rules. Human meetings consume reviewer capacity: one suggested policy reserves them for public, partner, or potentially external APIs, while allowing internal-only teams to opt in. section 5.5 section 5.5.1 section 5.5.10 section 5.5.2

Different ways to route a design

At Slack, a team shares an API definition with a cross-functional review channel; changes needing deeper discussion move to scheduled API office hours, and selected partners later review the design. Kubernetes instead defines mandatory and voluntary review categories, checks proposal prerequisites, and triages accepted requests on a board that shows status and reviewer load. Both make the path to feedback explicit without requiring every change to follow the same meeting format. section 5.6.3 section 5.6.4

Do not wait for a PR to be the first design conversation. Early guidance on the style guide, linters, API modeling, and service design can prevent problems that a late review would otherwise discover. Keep the review consumer-focused: start with what an API user sees, then work backward toward internal interfaces. section 5.3.3 section 5.7

How Automating API Delivery covers it

How deep it goes
Provides PR and meeting paths, a meeting sequence, review questions, and process-design goals; leaves exact governance rules to each organization.
Examples
Contrasts Slack’s review channel and office hours with Kubernetes’s mandatory-review criteria, triage board, and release-based prioritization.
What it assumes
Assumes an API definition change and a PR workflow; introduces the purpose of human review after automated checks.
What you can do after
Practice/perform a consumer-focused PR review and choose between asynchronous feedback and a meeting according to the change’s scope.
How it connects
Connects API style guides, linting, breaking-change checks, resource modeling, API documentation, and design-first development.

Also explained in

Compare how 2 books explain review process →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Automating API Delivery. The section links point to the parts of the book it draws on.