We build fast, scriptable CLIs for the infrastructure teams already depend on — Jenkins, Zabbix and Nexus — designed to be driven by AI coding agents like Claude Code as well as by engineers at the terminal.
# agent checks CI before shipping $ jk check [OK] Connect: https://jenkins.example.com (45ms) [OK] Auth: Logged in as 'alice' [OK] CSRF: Crumb issuer available Context is healthy and ready. # look at cluster health $ zx show_host_stats host1,host2 -d 7 # preview Docker tag cleanup (dry-run by default) $ nexus-cli docker cleanup docker-hosted --keep-last 10
AI agents are increasingly the ones running operational commands. Most DevOps tools were built for humans clicking through web UIs — not for an agent that needs predictable output, clear errors and guardrails.
JSON and CSV output, consistent exit codes and explicit health checks, so an agent can parse results instead of scraping screens.
Mutating commands run as a dry-run unless explicitly confirmed. Credentials stay in local profiles and never land in output or shell history.
Built-in rate limiting, retries with backoff and Retry-After handling, so automated workloads don't hammer shared controllers.
Open source under the MIT license. Each tool supports multiple environment profiles (dev, staging, prod) out of the box.
Jenkins CLI
config.xmlZabbix CLI & REPL
httptest)Sonatype Nexus 3 CLI
The same principles apply to every tool we ship.
Pure Go or a small Python package. Install it and run — no daemons or agents to deploy.
Read operations run directly; anything that changes state shows a preview first and needs an explicit flag.
When a command doesn't exist yet, a raw API fallback lets you — or your agent — call the underlying API with credentials injected.
Every tool started from day-to-day work running CI, monitoring and artifact infrastructure in production.
Questions, feedback, or want a CLI for another part of your DevOps stack? We'd like to hear from you.
founder@quangtd.cloud