jevci
Node quality gate that scores each diff, commit message and doc set with three Jev Score lenses and a security Noul.
Overview
JevCI is a quality gate for code changes that runs from the terminal, a git pre-commit hook, CI pipelines or a GitHub Action. It sends the diff, commit message and docs to Jev and asks four questions in parallel: three Score lenses for commit message quality (scope, rationale and conventional-commit style), documentation and code alignment (whether README and public API docs were updated when interfaces changed) and API contract and breaking-change risk, plus a Noul for secrets and security issues such as exposed keys or unsafe patterns.
The answers combine into a quality score from 0 to 100 with a Jev confidence, and the gate passes with exit code 0 or fails with exit code 1. The README claims sub-second end-to-end latency with about 100 to 250 ms of model inference, compared with 20 to 45 seconds for traditional LLM evaluations. It requires Node.js and a TypeSafe key.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How good is this commit message (scope, rationale, conventions)? | — |
| Score | Are docs aligned with the code changes? | — |
| Score | How risky is this change to the API contract? | — |
| Yes / no | Does this diff expose secrets or unsafe security patterns? | — |
How it works
- Collect diff, commit message and docs
- Ask four questions in one parallel Jev request
- Compute a 0-100 quality score with confidence
- Pass or fail with exit codes in hooks, CI or Actions
Strengths
- Fast enough for pre-commit hooks per the README
- Covers commits, docs, breaking changes and secrets
- Works locally and in CI
Caveats
- Speed claims are the author's
- Model-based secret detection should not replace dedicated scanners
At a glance
- Type
- CLI
- Status
- beta
- Language
- TypeScript
- Runtime
- Node CLI / GitHub Action
- Has code
- yes
Use cases and integrations
How curated lists describe it
Quality gate: asks four typed questions about each change — three `Score` lenses and one `Noul` — and blocks a diff, commit message or doc set that falls below the resulting quality score, from the terminal, a pre-commit hook or a GitHub Action.
yibie/awesome-jev
Frequently asked questions
What does JevCI check?
Commit message quality, doc and code alignment, breaking API risk and exposed secrets.
Where can it run?
Terminal, pre-commit hooks, CI pipelines and GitHub Actions.
How does it decide pass or fail?
It combines the answers into a 0-100 score and fails below the threshold.
Related projects
Advisor for coding agents: git and Stop hooks flag test tampering, secrets and skipped hooks, with optional Jev review findings.
HookbetaSemantic lint rules as yes/no questions to Jev, for review issues like logged secrets or unbounded retries that ESLint cannot express.
ProjectbetaTypeScriptGitHub Action that compiles .jev.yml rules into Jev questions and fails, comments on or passes a PR based on the numbers.
ProjectbetaTypeScriptRust code-review gate asking Jev small typed questions about functions and tests, failing CI only on rules measured 80% right.
CLIproductionRustCoding-agent hook that asks Jev one yes/no per project rule on every edit or turn and makes the agent fix any violation.
HookbetaTypeScriptRead-only Go CLI that checks files against your plain-language code rules with Jev, locally or in CI.
CLIbetaGo