code-review

name	code-review
description	Use this skill when the user asks for a code review. It automates checks and analysis.

Code Review

Overview

This skill provides a comprehensive code review process for the current branch. It verifies the code builds and passes tests across platforms, checks for performance regressions, and reviews the diff for bugs, improvements, and adherence to project standards.

Procedure

1. Preparation

1. Check Branch: Run git branch --show-current. If the output is main, inform the user that the current branch is main and reviews on main are not supported yet.
2. Check Remote: Run git fetch origin main. This ensures origin/main is up-to-date for accurate comparison.
3. Get Branch Point: Determine the merge base with git merge-base origin/main HEAD. Store this for later use.

2. Build & Test Verification

1. Build for All Platforms:
   • Run make build-all (covers Linux, Darwin, and Windows cross-compile, regardless of host OS).
   • Cleanup afterwards: make clean.
   • If any build fails, stop and report the error.
2. Lint & Format:
   • Run make fmt. If any files are modified, stop and report the error.
   • Run make fmt-md. If any files are modified, stop and report the error (Markdown is not formatted).
   • Run make vet. If any issues are found, stop and report the error.
3. Run All Tests:
   • Run make test to execute Go tests (unit + spec).
   • If tests fail, stop and report the errors.
4. Run WASI Tests:
   • Run make test-wasi to verify the WASI Preview 1 implementation.
   • If WASI tests fail, stop and report the errors.

3. Performance Verification

1. Compare Benchmarks: Run make bench-compare TARGET=. to compare performance against the main branch (the default base). NOTE: This can take several minutes as it runs the benchmark suite twice.
2. Review Results: Present the benchmark results to the user. Flag any regressions where:
   • Time (ns/op) increased by more than 5%.
   • Memory (B/op) increased.
   • Allocations (allocs/op) increased.

4. Dependency Check

1. Detect New Dependencies: Run git diff origin/main..HEAD -- go.mod go.sum.
2. Verify Module Consistency:
   • Run go mod tidy.
   • Run git diff --name-only go.mod go.sum.
   • If there are changes, stop and report that go mod tidy needs to be run.
3. Report Changes: If there are any changes to go.mod or go.sum:
   • CRITICAL: This project is pure Go with no CGo. New dependencies are strongly discouraged.
   • Report exactly what dependencies were added or modified.
   • Ask the user to confirm the dependency changes are intentional.

5. License Header Check

1. Find New Files: Run git diff --name-status origin/main..HEAD | grep '^A' to list all newly added files.

2. Check Headers: For each new source file (.go, .py, .yml, etc.), verify it contains the proper license header:

   // Copyright [Year] Google LLC
   //
   // Licensed under the Apache License, Version 2.0 (the "License");
   // you may not use this file except in compliance with the License.
   // You may obtain a copy of the License at
   //
   //     http://www.apache.org/licenses/LICENSE-2.0
   //
   // Unless required by applicable law or agreed to in writing, software
   // distributed under the License is distributed on an "AS IS" BASIS,
   // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
   // See the License for the specific language governing permissions and
   // limitations under the License.
   

   Use the appropriate comment syntax for the file type (e.g., # for Python and YAML).

3. Report Missing Headers: List any files missing the license header.

6. Code Review

1. Get the Diff: Run git diff origin/main..HEAD to see all changes.

2. Review for Issues: Analyze the diff and check for:

   Bugs & Correctness:

   • Logic errors and edge cases.
   • Improper error handling (errors must be handled or propagated, never ignored).
   • Data races or concurrency issues.
   • Off-by-one errors.

   Security & Sandboxing: Treat untrusted Wasm and host inputs as hostile. The bullets below are a floor, not a ceiling — hunt for classes of flaw they don't name.

   • Validator is the boundary: the VM trusts it, so any skipped or weakened check there is exploitable.
   • Untrusted Wasm or invalid host inputs must never panic — failures surface as standard Go errors.
   • Sandbox escapes: untrusted code reaching host state it shouldn't see.

   Performance (especially in hot paths like the VM loop):

   • Unnecessary heap allocations.
   • Inefficient algorithms.
   • Missed opportunities to reuse buffers/slices.
   • Unnecessary copying of data.

   Simplicity & Readability:

   • Overly complex code that could be simplified.
   • Duplicate code that could be refactored.
   • Poor variable naming (names should be self-explanatory).
   • Unnecessary comments (code should be self-explanatory). Verify no "TODO", "FIXME", or placeholder comments were left behind.

   Documentation:

   • Ensure README.md and other documentation is updated if new features, flags, or configuration options were added.
   • Ensure AGENTS.md is still accurate after the change: if the diff alters the project layout, build/test commands, conventions, or constraints it documents, it must be updated to match.

   Modern Go Features:

   • Opportunities to use newer Go features (e.g., generics, improved slices package, range-over-int, etc.).
   • Deprecated patterns that should be updated.

   Project Standards:

   • Adherence to rules in AGENTS.md.
   • Consistent style with existing code.

3. Compile Findings: Create a structured report with:

   • Critical Issues: Bugs, correctness problems, or security concerns that must be fixed.
   • Suggestions: Performance improvements, simplifications, or style enhancements.
   • Notes: Minor observations or questions.

7. Report

Present a summary to the user containing:

1. ✅ or ❌ for each verification step (build, lint, tests, spec tests, WASI tests).
2. Benchmark comparison results with any regressions highlighted.
3. Dependency change report (if any).
4. List of files missing license headers (if any).
5. Code review findings organized by severity.
6. An overall recommendation: Ready to Merge or Needs Changes.