Jev-Code-Reviewer skill
Jev-Code-Reviewer ranks changes in an agent PR as P0, P1 or P2 and retells them in plain language on the GitHub page.
- 152 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Changed code and context are sent to TypeSafe and OpenAI.
- Keys are stored in an unencrypted file.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/jev-code-reviewerDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Changed code and context are sent to TypeSafe and OpenAI.
- Keys are stored in an unencrypted file.
This is third-party code. Review the repository files before installing.
What it does
Jev-Code-Reviewer, a local review tool, cuts the effort of reading large agent-generated PRs. Every change gets a P0, P1 or P2 priority and only P0 opens by default. The diff is retold in plain language and the original code is one toggle away. The `jev-reviewer analyze --pr <URL> --repo <path>` command classifies the changes and a Chrome extension replaces code on the Files changed page. The `jev-reviewer` skill asks an agent to run analysis after opening a PR. The tool runs on the user's machine and posts nothing to GitHub.
Who it is for. Developers whose agents open large PRs and tech leads who want to see the most important parts of a change first.
Good fit when
- An agent finished a task and opened a PR across dozens of files.
- A reviewer reads P0 first and leaves the rest collapsed.
- An explanation of the logic of a change is needed without reading every hunk.
- Priorities are tuned for a repository through `.jev-reviewer.json`.
Not a fit when
- Code must not leave the machine, since analysis goes through TypeSafe and OpenAI.
- Review must be posted as GitHub comments, and the tool posts nothing.
- The GitHub page uses the new `/changes` interface, which is unsupported.
Example request
Analyse the open PR with `jev-reviewer analyze` and list the P0 changes with a short explanation each.Limitations
Analysis sends changed code and context to TypeSafe and OpenAI, and both keys are required. By default the first 12 change units are analysed and `--max-units` raises the cap to 100. Keys sit unencrypted in `~/.config/jev-reviewer/credentials.json`. Only the classic Files changed page is supported, and old reports are hidden after new commits. Node.js 22, GitHub CLI and a repository clone are required.
How to disable. Remove the skill from `~/.claude/skills/jev-reviewer` or `~/.agents/skills/jev-reviewer`, the Chrome extension and the `jev-reviewer` command. Delete the credentials file by hand.
FAQ
Does it post comments to GitHub?
No. It runs locally and only changes what you see on the Files changed page in your browser.
What does a priority mean?
Attention, not correctness. P0 opens by default, P1 and P2 start collapsed, and thresholds are tuned in `.jev-reviewer.json` or with `--policy`.
Where does the code go?
Changed code and nearby context go to TypeSafe for classification and OpenAI for explanations. The extension talks only to the local server `127.0.0.1:4731`.
How can I try it without keys?
Run `npm run demo` and open `http://127.0.0.1:4731/demo`. The replay calls no providers.
Does the install script add the CLI?
No. `install/install-agent-skill.sh` installs only the skill; the CLI is linked with `npm link`.
Discussion
No questions or reviews yet.
Related
SkillsA plugin with a local MCP server: the agent scores its own code on a set of criteria through Jev and improves it by the numbers
A code-review CLI and plugin for agents: reads a git diff and writes line-level comments with context
A CLI and three skills that work through review-bot comments on a PR: assess findings, fix code, reply and wait for checks
reviewd reviews GitHub, GitLab and BitBucket pull requests through Claude Code, Gemini or Codex CLI and comments on them
CLI with a virtual board of AI experts that reviews a PR or local diff and returns a score, severity level and evidence.
Two skills: a pull request review where two models argue, and a helper that works through review comments round by round
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Jev-Code-Reviewer (this card) | no | 152 | |
| Human Review | no | 1,454 | a manual review workflow; Jev-Code-Reviewer prepares priorities and a retelling of changes for the reviewer |
| code-review-skill | no | 2,256 | gives the agent a review checklist; here the result is shown to a human on the GitHub page |
README in short
Classification works on change units, not files. The report shows P0 items and a retelling of the logic, while files with unanalysed changes keep GitHub's own code.