Jev directory
CommunityJev Tools & CLIs
Jev Developer Harness
An open-source CLI harness for AI coding agents such as Codex, Claude Code, Cursor, and Aider that uses TypeSafe AI Jev for context ranking, command guarding, patch review, and semantic linting.
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- Commands include context rank, guard check, patch review, hooks install, lint semantic.
- Guard check blocks destructive commands like git reset --hard.
- Patch review reports regression risk, scope, auth, and secret exposure.
- Git pre-commit hook aborts commits when secrets or regressions are detected.
- Benchmark table on the page is labeled hypothetical, not measured.
Questions about this entry
Short answers drawn from the linked project or page.
Run npx -y jev-dev compare to execute the terminal comparison simulation with zero install. The full harness commands are available once the package is installed.
The package describes itself as a runtime safety and intelligence harness for AI coding agents including Antigravity, Codex, Claude Code, Cursor, VSCode, and Aider.
jev-dev guard check evaluates a shell command and returns ALLOWED or BLOCKED. Destructive commands such as git reset --hard HEAD~1 are blocked, and network operations like git push need an explicit --allow-network flag.
It audits a diff or commit range against a task description and prints judgments such as regression risk, out-of-scope percentage, auth and database modification, missing tests, secret exposure, and a confidence figure.
After jev-dev hooks install, each git commit evaluates the staged diff. If secrets, private keys, or catastrophic regressions are detected the commit aborts with exit 1; otherwise it proceeds. jev-dev hooks uninstall removes it.
No. The page explicitly labels the comparison values as hypothetical assumptions, not measured benchmarks, and states no valid token or cost baseline has been collected.
The sample GitHub Actions workflow passes TYPESAFE_API_KEY and OPENROUTER_API_KEY as repository secrets and runs npx jev-dev lint semantic over the pull request commit range, optionally in advisory mode.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: A usage router for Grok Bot
X: A prompt box that fills itself in
X: Vercel's fx safety reviewer, measured at 18× faster
X: How Jev makes agents faster and cheaper
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.