Practical guidance for reviewing agent-generated work.
Use concrete workflows for reviewing implementation plans, rendered documents, product decisions, and preview applications while keeping final acceptance human-owned.
Start with the flagship argument or choose the guide closest to the work in front of you.
Start with the review decision you need to make.
Each guide is maintained as indexable, server-rendered content with a practical workflow and related product path.
How to review an AI-generated implementation plan
Challenge architecture, failure states, migration, testing, rollout, and ownership before coding starts.
Read the guideHow to review Markdown pull requests as rendered documents
Read the complete artifact first, then use source diffs for precise verification.
Read the guideA practical human-in-the-loop workflow for coding agents
Define the artifact, feedback, permission, revision, verification, and approval stages.
Read the guideFollow the artifact, team, or integration path.
Blog updates remain artifact-backed. Evergreen guides and the flagship article live here in the application repository.
Agent workflows
Plan review, agent integrations, structured revision context, and human approval.
ExploreDocument review
Rendered Markdown, HTML, semantic comments, and complementary source diffs.
ExploreProduct and engineering decisions
ADRs, specifications, rollout plans, acceptance criteria, and forms.
ExploreLive application review
Selected UI context from opted-in preview applications.
ExploreProduct updates
Artifact-backed field notes from the Commentary blog.
ExploreResearch
Human-owned studies, evidence, and review-informed product decisions.
ExploreDiffs Are Not Documents
Why agent-generated plans, specifications, and documentation need a review surface designed for meaning—not only changed lines.
Review the next agent-generated artifact.
Try a plan review, open a public GitHub review, or connect the coding agent your team already uses.