Description
review-loop is an agent skill that runs an iterative worker and reviewer cycle inside a single session. The agent does the work, spawns a reviewer subagent that critiques it and scores it from 1 to 10 with actionable feedback, then revises the work based on that feedback, and repeats until a quality gate is met. Its core principle is that first drafts are never final, and that repeated critique tends to produce better output than a single pass. The skill works best with agents that can spawn subagents. Where that is not available, it describes a way to simulate the reviewer step by opening a fresh chat context and pasting in only the work to be reviewed. It is useful for drafts that matter, such as documents, code or designs, where a second, independent look is worth the extra cycle. The skill is listed on skills.sh under the 2dmurali review-loop-skill repository and is free.…
Highlights
- Worker and reviewer cycle within one session
- Reviewer subagent scores work from 1 to 10
- Revises and repeats until a quality gate is met
- Fallback for agents without subagent support
Ratings & Reviews
No ratings yet
Be the first to rate review-loop.
People also visit
review-loop FAQ
What does review-loop do?
It makes the agent review its own work through a separate reviewer, then revise until the quality gate is met.
How is the work scored?
A reviewer subagent scores it from 1 to 10 and gives actionable feedback.
What if my agent cannot spawn subagents?
The skill describes simulating the reviewer in a fresh chat context, pasting only the work.
Who is it for?
Anyone whose agent drafts documents, code or designs that benefit from a second critique.
How do I install it?
Run npx skills add https://github.com/2dmurali/review-loop-skill --skill review-loop in your terminal.
Is it free?
The listing shows the skill as free.
What are the best review-loop alternatives?
Popular alternatives to review-loop include bmad-code-review, chief-of-staff and agent-self-evaluation. See all review-loop alternatives