Skip to content

Review Agents

Review agents are specialized checklists you configure in the ThinkReview portal. Each agent has a goal, scope, and optional references (for example OWASP guidance or internal secure-coding standards). Agents run alongside the main review so you get niche coverage without a second workflow.

Manage agents at portal.thinkreview.dev/agents.

  • Security-focused reviews grounded in references your team trusts
  • Migration, performance, or framework-specific checks
  • Separate risk profiles for different teams (ThinkReview supports up to ten agents)

Agent reviews run in parallel with the main analysis. While agents are running, you can close the panel and return shortly—work continues in the cloud.

  1. Sign in to the ThinkReview portal.
  2. Create an agent: set the goal, scope, and optional reference material.
  3. Enable the agent so it runs with your normal ThinkReview reviews.
  4. Open a pull or merge request, run ThinkReview, and open the agent tabs for specialized findings.

ThinkReview can also suggest niche agents tailored to projects you review, based on stack signals—ready to enable when they fit.

  • Agents are centered in the extension and portal. ThinkReview MCP returns the main structured review and does not run custom agents by default.
  • Combine agents with Custom Rules for global standards plus specialized checklists.
  • For a walkthrough focused on security, see the blog post Security agents workflow.

TL;DR: Review agents are specialized, reference-backed checklists (up to ten) that run with every review from the portal and extension sidebar.