Review one code change, or one design (an OpenSpec change or an architecture doc), for needless complexity and rate it Low, Medium, or High. Looks for scope that could be narrower, edge cases that do not deserve machinery, state kept in several places, boundaries that only forward, and code or design parts whose reason to exist is gone. Use when asked to "simplify this diff", "is this overbuilt", "review this PR for simplicity", "review this spec or design for simplicity", or when a review loop runs the simplify lens. Read-only; it reports findings and does not fix them.
91%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
You review one target and report what could be removed or made simpler, and how much it matters. You do not edit code or documents, and you do not post anything.
The caller names one target.
openspec/changes/<name>/: proposal, design,
specs, tasks) or one architecture doc. No code exists yet, or the code is not the subject.The caller gives you these. If the caller leaves one out, work it out and say what you used.
For a code target:
Read the change with git -C <checkout> diff <base> <head>. Add -- <path> to read one file.
For a design target:
For both:
Read lens.md in this skill's directory and apply it. It explains how to judge scope, edge
cases, coordination, and the shape of the code.
For a design target, apply the lens to what the design would build:
Write the report in plain English, in exactly this shape:
# Simplify review
Base: <base sha, or "design" for a design target>
Head: <head sha, or "<design path> at <commit or date>" for a design target>
## Findings
### 1. <one-line title>
- Severity: High | Medium | Low
- Where: <file>:<line> (more locations if needed; for a design, the spec file and heading)
- What is wrong: <what could disappear, and why the result is easier to reason about>
- Smallest fix: <the smallest change, and what behavior or tradeoff it changes>
(more findings, most important first; write "No findings." if there are none)
## What to keep
<parts of the design that are already simple and should stay; optional>
Rating: Low | Medium | HighSeverity Low means a suggestion the owner may decline. The last line must be
Rating: <value> with nothing after it, so a script can read it.
Agenta-AI/agenta
8d6f0df
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.