Process

Automated review

You might say: agent review, second opinion

A separate run checks a proposed change with fresh context and reports findings before human review.

What it does
Adds a second, independent judgement, so the run that wrote the change is not the only one that checks it.
Ask
Is the reviewer separate enough to catch what the author missed, and does it report evidence for each finding?

What to look for

  1. A review check authored by a different bot or run
  2. Findings linked to changed lines with severity and evidence
  3. A second review status beside deterministic checks

Connections

Sources

  1. Anthropic, Best practices for Claude Code, Give Claude a way to verify its worka verification subagent or a dynamic workflow that checks its own findings has a fresh model try to refute the result
  2. Addy Osmani, Agentic Code Review, The tools are good, but not always for the reason they advertiseA second model with different priors genuinely catches real bugs, and it catches a lot of them
  3. Titus Winters et al., Software Engineering at Google, CritiqueUploading a snapshot of the change triggers code analyzers
  4. Anthropic, Building effective AI agents, Agents in practiceHowever, whereas automated testing helps verify functionality, human review remains crucial for ensuring solutions align with broader system requirements.
  5. Gene Kim and Steve Yegge, Vibe Coding: Building Production-Grade Software With GenAI, Chat, Agents, and Beyond, From Line Cook to Head Chef: Orchestrating AI Teams (book excerpt)Generators and verifiers: Separate concerns by creating dedicated agents for implementation versus testing.
  6. Lauren Tan, pstack, When the night holds a queue, not a taskEach PR gets one owner agent that carries it from build through merge, and no owner merges on its own verdict. A swarm of fresh verifiers checks the owner's code-ready head.
  7. Lauren Tan, pstack, Land the stack with Shippingthe agent that judges a change is never the one that wrote it.
  8. Matt Pocock, software-factory skill (pull request #1316), Start slowa separate run that reads the diff cold against the codebase's standards and the ticket, when there is one.
  9. Matt Pocock, AI Coding Dictionary, Automated reviewAn agent reviewing another agent's work, often with a different model or system prompt. Non-deterministic: it forms a judgement.
  10. Matt Pocock, AI Coding Dictionary, Automated reviewA reviewer with a fresh context window has none of that attachment: it sees the diff the way a stranger would, which is what review depends on.
  11. Matt Pocock, AI Coding Dictionary, SycophancyBiased framing — review skews positive when you signal you wrote it; negative when you signal someone else did. Same artifact, different verdict.

Use it on your project

teardown-apply: hold "Automated review" (automated-review) from the teardown at https://teardownatlas.com/software-factory/ against my project

Paste it into a coding agent that has the teardown skills.

An AI agent wrote this map with the teardown skill. It can contain errors: check the sources before you rely on a part. About and legal