- Home
- Skills
- Hikaruegashira
- Agent Skills
- Review Flow
review-flow_skill
- Python
1
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill hikaruegashira/agent-skills --skill review-flow- SKILL.md1.3 KB
Overview
This skill runs autonomously after a pull request is created to act as a paired reviewer and session facilitator on behalf of the PR author. It switches to the PR worktree, runs parallel subagents to validate behavior, trim unnecessary code, and fix CI issues, then produces a concise report and starter questions for a pair-programming session. The goal is to make the PR easy for human reviewers to understand and actionable to merge.
How this skill works
On trigger (review / pr created / pr open) the skill checks out the target worktree with the command gh wt co $1 and ensures you are in the correct repository path. It launches three subagents in parallel: QA to run realistic usage checks, deslop to remove dead or unnecessary code and push changes, and fixci to address CI failures and push fixes. Finally it aggregates results into a codemap, try instructions, and a set of guiding session questions.
When to use it
- Immediately after opening a pull request to prepare it for reviewers
- When CI is flaky or failing and you want automated remediation attempts
- When you want a quick runtime validation of the change with realistic scenarios
- When cleaning up unused code or simplifying changes before review
- When you want starter questions to drive a focused pair-programming review
Best practices
- Run from the repository root or move to ~/ghq/github.com/{owner}/{repo} before invoking worktree switch
- Use the exact PR identifier with gh wt co $1 to avoid checking out the wrong branch
- Review subagent outputs before merging pushes made by deslop and fixci
- Keep QA scenarios concise and representative of real user flows
- Treat the generated session questions as prompts, not exhaustive test coverage
Example use cases
- A feature PR is opened; QA validates baseline flows, deslop removes experimental flags, fixci repairs a broken pipeline step, and the skill reports a clean codemap.
- A bugfix PR triggers the skill; QA reproduces the bug with expect scripts, deslop removes outdated debug code, and fixci updates CI config to restore green builds.
- A refactor PR runs the skill to produce a codemap that explains moved functions and to generate pair-session prompts for architecture review.
FAQ
A concise explanation of modified files, key functions, and how the changes affect behavior or data flow.
How are QA checks specified?
QA runs realistic use-case scripts; the skill outputs shell + expect steps under the try section you can replay locally.